General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

Users may download and print one copy of any publication from the public portal for the purpose of private study or research.

You may not further distribute the material or use it for any profit-making activity or commercial gain

You may freely distribute the URL identifying the publication in the public portal If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from orbit.dtu.dk on: Jan 09, 2021

Benchmarking of Processes for the Biosynthesis of Natural Products

Seita, Catarina Sanches

Publication date:2016

Document VersionPublisher's PDF, also known as Version of record

Link back to DTU Orbit

Citation (APA):Seita, C. S. (2016). Benchmarking of Processes for the Biosynthesis of Natural Products. Technical University ofDenmark.

BENCHMARKING OF PROCESSES FOR THE BIOSYNTHESIS OF NATURAL PRODUCTS

Catarina Sanches SeitaPhD Thesis

November 2016

Benchmarking of Processes for the

Biosynthesis of Natural Products

PhD Thesis

Catarina Sanches Seita

Department of Chemical and Biochemical Engineering

Technical University of Denmark

November 2016

Copyright©: Søren Heintz

Address:

Phone:

Web:

September 2015

Process and Systems Engineering Center (PROSYS)

Department of Chemical and Biochemical Engineering

Technical University of Denmark

Building 229

Dk-2800 Kgs. Lyngby

Denmark

+45 4525 2800

www.kt.dtu.dk/forskning/prosys

Print: STEP

Catarina Sanches Seita

November 2016

i

Abstract

Natural products constitute an extensive family of organic molecules with more than 200,000 compounds

discovered in several natural sources (plants and microbes). Most of these compounds have a very complex

structure, multiple chiral centers and can have different biological activities. These biological activities can

be of interest for use in different sectors of chemical industry, in particular pharmaceutical industry where

several drugs are derived or inspired by natural products structure.

However, the large scale production of natural products is hindered by its relatively poor abundance in

nature, which makes extraction from natural sources an economically unfeasible technology in most of the

cases. Chemical synthesis is also very difficult given the structural complexity and chirality of the molecules.

Synthetic biology offers very promising tools for production of natural products. Genetic engineering allows

expressing the genes responsible for the biosynthesis of these natural products and to insert them into

fermentable organisms like yeast or bacteria. And protein engineering offers the possibility to engineer the

enzymes that perform the natural products’ biosynthesis, which allows the production of these complex

molecules in single or multi-steps biocatalytic reactions.

In this thesis, a systematic approach for route selection and screening of the different process options to

manufacture a natural product is presented. This methodological approach includes a set of evaluation tools

to assess processes both from an economic and environmental perspectives and it is demonstrated with two

case-studies. For each case-study different tools are used to evaluate the process.

The first case-study consists of the bioconversion of (R)-limonene to (R)-perillic acid by Pseudomonas putida

GS1. (R)-perillic acid is a monoterpenoic acid with antimicrobial properties. It has a strong inhibitory effect

on bacteria and fungus, which makes it an attractive compound to be used as a preservative for instance in

cosmetic industry, but on the other hand makes the biosynthesis a complicated process to develop. An

environmental assessment of the different synthetic routes for (R)-perillic acid production showed that

biosynthesis represents the most promising option. This process was further explored through an economic

assessment and process modelling including a sensitivity analysis on key process metrics, which allowed the

identification of the main process bottlenecks. Product inhibition and substrate loss were identified as some

of the main process limitations and strategies for improving them were suggested.

The second case-study describes the production of a recombinant protein, brazzein, by fermentation of the

yeast Pichia pastoris. Brazzein is a natural sweetener found in the fruits of the African plant Pentadiplandra

ii

brazzena Baillon, which is sweeter than regular sucrose by several orders of magnitude. Here different tools

were applied to evaluate the environmental profile of the process in comparison with other sweeteners.

The main benefit of this early-stage evaluation is putting the biosynthesis of natural products into context in

relation to demands of an industrially feasible chemical process. Moreover, it can give very meaningful insight

into process development and provides a good overview of the whole reaction and process. The proposed in

silico approach can guide research and development and ultimately contribute to the implementation of

more bioprocesses for the production of natural products.

iii

Resumé

Naturlige produkter er en stor familie af organisk molekyler der indeholder mere end 200.000 forbindelser

opdaget i natur (planter og mikroorganismer). De fleste af disse forbindelser har en kompleks struktur og

flere kiralecentre og kan have forskellige biologiske aktiviteter. Disse biologiske aktiviteter kan være

interessante for forskellige sektorer af kemi industri, især for den farmaceutiske industri hvor mange

lægemidler er afledt eller inspireret as naturlige produkters struktur.

Industriel produktion af naturlige produkter er begrænset af deres lave koncentrationer i naturen, der gør

ekstraktion en økonomisk umulig teknologi. Kemisk syntese er også en meget vanskelig opgave på grund af

den strukturelle kompleksitet og den molekylære asymmetri.

Syntetisk biologi tilbyder mange lovende værktøjer til produktion af naturlige produkter. Genteknologi gør

det muligt at udtrykke generne der er ansvarlig for biosyntese af disse naturlige produkter og indsætter dem

i fermenterbare mikroorganismer, som gær og bakterier. Og proteinoptimering gør det også muligt at

syntetisere naturlige produkter i enkelt- eller flertrins biokatalytiske reaktioner.

Arbejdet i denne afhandling præsenterer en systematisk metode til udvælgelse af ruter og processer til

produktion af naturlige stoffer. Denne metodiske fremgangsmåde inkuderer værktøjer for økonomisk og

miljømæssig evaluering af bioprocesser og er blevet valideret gennem to forskellige eksempler.

Det første eksempel er produktionen af (R)-perillic acid fra (R)-limonene ved brug af Pseudomonas putida

GS1. (R)-perillic acid er en monoterpen syre med antibakteriel aktivitet. Det er interessant at bruge (R)-perillic

acid som en konserveringsmiddel i den kosmetiske industri, men dens biosyntese er en vanskelig opgave. En

miljømæssig evaluering af forskellige processer til produktion af (R)-perillic acid viste at biosyntesen er det

mest lovende alternativ. De økonomiske aspekter af processen blev også undersøgt baseret på en

procesmodel der inkluderer en sensitivitetsanalyse af de væsentlige parametre. Dette muliggjorde

identifikationen af de primære flaskehalse i processen. Produktinhibering og substrattab blev identificeret

som de største begrænsninger for processen, og en række strategier blev foreslået for at udbedre disse.

Det andet eksempel er produktionen af det rekombinante protein, brazzein, ved fermentering af gæren

Pichia pastoris. Brazzein er et naturligt sødemiddel som findes i frugterne af den afrikansk plante

Pentadiplandra brazzena Baillon, der er adskillige gange sødere end sukrose. I dette eksempel blev en

miljømæssig evaluering anvendt til at sammenligne forskellige processer for sødemiddel.

iv

Den største fordel ved denne type evaluering er at kunne sætte kravene for en industriel biosyntese af

naturlige produkter i perspektiv. Den kan også give meningsfuld indsigt i procesudviklingen og et overblik

over reaktionssystemet og processen. Denne in silico fremgangsmåde kan guide procesudviklingen og

bidrage til introduktionen af flere bioprocesser i produktionen af naturlige produkter.

v

Preface

The work presented in this thesis was conducted in CAPEC-PROCESS research center at the Department of

Chemical and Biochemical Engineering from the Technical University of Denmark as partial fulfillment of the

requirements for acquiring a PhD degree in engineering. The work was developed in the period from

December 2013 until November 2016 and was supervised by Professor John M. Woodley (DTU Chemical

Engineering) and Dr. Mathias Nordblad (DTU Chemical Engineering) and Dr. Jessica Rehdorf (BRAIN AG) as

co-supervisors.

The work has been performed within the strategic alliance NatLifE 2020 coordinated by BRAIN (Zwingenberg,

Germany) and funded by the German Federal Ministry of Education and Research (BMBF) (grant no.

031A206-B).

Kgs. Lyngby

November 2016

Catarina Seita

vi

vii

Acknowledgments

I acknowledge my supervisors, John M. Woodley, Mathias Nordblad and Jessica Rehdorf, for their valuable

contributions and the opportunity to work on this project. Thank you John for giving me the freedom that

allowed me to grow so much as a scientist and as a person. Thank you for all the scientific discussions and

for pushing me to think “out of the box”. Thank you Mathias for co-supervising me and thank you Jessica for

always being available for meetings and discussions about the project.

I would also like to thank Christian Naumer for the support in the work for brazzein case-study.

I wish also to acknowledge all my former students who contributed to this project, Julia Wagner, Rowan

Lindeque, David Veiga and Julia Keim.

I am also deeply grateful to all my colleagues in CAPEC-PROCESS for the companionship and for creating a

nice working environment. A special thanks to Asbjørn, Francesco and Murray for all the help with the

laboratory work. And finally, thanks to my office mates Rohana and Maria for all the fruitful discussions,

motivation and for being such a great company.

Last but not least, thank you to my family, friends and to João for all the support and patience throughout

the project.

viii

ix

List of abbreviations

Nomenclature Description

ACS American Chemical Society API Active pharmaceutical ingredient

CCM Corn cob mix ChemSec International Chemical Secretariat

DMAc N,N-dimethylacetamide DMF Dimethylformamide

DMSO Dimethyl sulfoxide D-PheOMe D-phenylalanine methyl ester

DSP Downstream processing EHS Environmental, health and safety FCI Fixed-capital investment

GCIPR Green Chemistry Institute’s Pharmaceutical Roundtable GMO Genetically modified organism GSH Globally harmonized system of classification and labelling of chemicals HPLC High-performance liquid chromatography ISPR In situ product removal ISSS In situ substrate supply

KREDs Ketoreductases LCA Life cycle assessment LCI Life cycle inventory

LCIA Life cycle impact assessment L-PheOMe L-phenylalanine methyl ester

MTBE Methyl tert-butyl ether NMP N-methylpyrrolidinone PM Phenylalanine methyl ester

RAPEX Rapid alert system for dangerous non-food products REACH Registration, evaluation, authorization and restriction of chemicals

SE Sucrose equivalent SIN Substitute it now STY Space-time yield TB Terrific broth

UPS Upstream processing ZA N-benzyloxycarbonyl-aspartic acid

WFI Water for injection

x

xi

List of nomenclature

Nomenclature Description Unit

AP Acidification potential g SO2-eq AE Atom economy %

CAPEX Capital cost € cdw Cell dry weight g CE Carbon efficiency % ee Enantiomeric excess %

E-factor Environmental factor kg waste/kg product EMY Effective mass yield % EP Nutrient enrichment potential g PO4

3--eq ETP Eco-toxicity potential PAF.m3.day

GWP Global warming potential g CO2-eq HTP Human-toxicity potential CTUh HWP Hazardous waste potential g waste MW Molecular weight g/mol

OD600nm Optical density at 600 nm Dimensionless OPEX Operational cost € pKa Acid disassociation constant Dimensionless PMI Process mass intensity kg materials/kg product

POCP Photochemical ozone creation potential g C2H4-eq RME Reaction mass efficiency %

SI Solvent intensity kg solvents/kg product SOP Stratospheric ozone depletion potential g CF11-eq

V Volume L WI Water intensity kg water/kg product YSX Yield coefficient of biomass on substrate kg biomass/kg substrate

xii

xiii

Table of contents

Abstract ............................................................................................................................................................... i

Resumé .............................................................................................................................................................. iii

Preface ................................................................................................................................................................ v

Acknowledgments ............................................................................................................................................ vii

List of abbreviations .......................................................................................................................................... ix

List of nomenclature .......................................................................................................................................... xi

Table of contents ............................................................................................................................................. xiii

1. Introduction ............................................................................................................................................... 1

1.1 Background ........................................................................................................................................ 1

1.2 Bioprocesses ...................................................................................................................................... 3

1.2.1 Biocatalytic processes ................................................................................................................ 3

1.2.2 Fermentation processes ............................................................................................................ 6

1.3 Natural products ................................................................................................................................ 8

1.4 Scope of the thesis .......................................................................................................................... 12

1.5 Thesis outline ................................................................................................................................... 13

2. Sustainability assessment of bioprocesses .............................................................................................. 17

2.1 Economic assessment ...................................................................................................................... 18

2.1.1 Process metrics ............................................................................................................................ 18

2.1.2 Full cost estimation...................................................................................................................... 20

2.2 Environmental assessment .............................................................................................................. 23

2.2.1 Green Chemistry .......................................................................................................................... 25

2.2.1.1 The natural fit of biocatalysis with green chemistry ................................................................... 27

2.2.1.2 Green Chemistry metrics ............................................................................................................. 30

Reaction-related metrics ......................................................................................................................... 31

xiv

Process-related metrics ........................................................................................................................... 33

2.2.2 Environmental, Health and Safety hazards ................................................................................. 40

2.2.3 Life Cycle Assessment .................................................................................................................. 41

2.3 Concluding remarks ......................................................................................................................... 43

3. Case-study I: Bioconversion of (R)-limonene to (R)-perillic acid by a whole-cell biocatalyst ................. 47

3.1 Introduction ..................................................................................................................................... 47

3.2 Bioprocess description .................................................................................................................... 47

3.3 Benchmarking of different routes for perillic acid production........................................................ 49

3.4 Bioprocess evaluation...................................................................................................................... 53

3.4.1 Economic assessment .................................................................................................................. 53

3.4.1.1 Process metrics ............................................................................................................................ 53

3.4.1.2 Process model .............................................................................................................................. 55

3.4.2 Environmental assessment .......................................................................................................... 60

3.5 Experimental characterization of the bioprocess............................................................................ 62

3.5.1 Materials and methods................................................................................................................ 62

3.5.1.1 Determination of the reaction rate ......................................................................................... 62

3.5.1.2 Measurement of substrate loss under fermentation conditions ............................................ 64

3.5.2 Results ......................................................................................................................................... 65

3.5.2.1 Determination of the reaction rate ......................................................................................... 65

3.5.2.2 Measurement of substrate loss under fermentation conditions ............................................ 68

3.6 Strategies for improvement of the bioprocess ............................................................................... 70

3.6.1 Materials and methods................................................................................................................ 72

3.6.1.1 High-cell-density fermentation................................................................................................ 72

3.6.2 Results ......................................................................................................................................... 73

3.6.2.1 High-cell-density fermentation................................................................................................ 73

3.7 Concluding remarks ............................................................................................................................... 75

4. Case-study II: Fermentation of recombinant Pichia pastoris for brazzein production ........................... 77

xv

4.1 Introduction ..................................................................................................................................... 77

4.2 Sweeteners ...................................................................................................................................... 77

4.2.1 Intensity of sweetness ............................................................................................................. 81

4.3 Bioprocess description .................................................................................................................... 82

4.4 Bioprocess evaluation...................................................................................................................... 85

4.5 Environmental assessment .............................................................................................................. 86

4.5.1 Scope of the environmental assessments ............................................................................... 87

4.5.2 Modeling and inventory of the processes ............................................................................... 88

4.5.2.1 Brazzein extraction from genetically modified maize ............................................................. 88

4.5.2.2 Brazzein extraction from the African fruits of Pentadiplandra brazzeana Baillon .................. 92

4.5.2.3 Stevia extraction from leaves .................................................................................................. 94

4.5.2.4 Aspartame enzymatic production ........................................................................................... 97

4.5.3 Benchmarking of brazzein fermentation process with similar processes ................................... 99

4.5.3.1 Process Mass Intensity ............................................................................................................ 99

4.5.3.2 Product concentration ........................................................................................................... 103

4.5.3.3 Water Intensity ...................................................................................................................... 104

4.5.3.4 E-factor .................................................................................................................................. 105

4.5.3.5 Origin of the raw materials .................................................................................................... 105

4.5.4 Environmental, health and safety aspects of P. pastoris fermentation .................................... 108

4.6 Concluding remarks ....................................................................................................................... 109

5. Discussion .............................................................................................................................................. 115

5.1 Methodology for development of processes for the biosynthesis of natural products ............... 115

5.2 Case-study I: Bioconversion of (R)-limonene to (R)-perillic acid by a whole-cell biocatalyst ....... 119

5.3 Case-study II: Fermentation of recombinant Pichia pastoris for brazzein production ................. 121

5.4 General recommendations for implementation of processes for the biosynthesis of natural

products ..................................................................................................................................................... 123

6. Conclusions ............................................................................................................................................ 125

xvi

7. Future work ........................................................................................................................................... 127

8. References ............................................................................................................................................. 129

Appendices .......................................................................................................................................................... i

Mass balances for the chemical route for (R)-perillic acid synthesis from β-pinene .................................. i

Calibration curve for HPLC measurements of (R)-perillic acid .................................................................. iii

Calibration curve for the UV-VIS measurements of (R)-limonene at 254 nm ........................................... iv

Mass balances for the processes assessed in case-study II ........................................................................ v

1. Brazzein extraction from the fruits of Pentadiplandra brazzeana Baillon ......................................... v

2. Brazzein fermentation ....................................................................................................................... vi

3. Brazzein extraction from GMO maize ............................................................................................... ix

4. Stevia production .............................................................................................................................. xi

5. Aspartame production ..................................................................................................................... xiii

1

1. Introduction

This thesis attempts to create a bridge between science, engineering and economy and apply different

concepts of these areas to the development of processes for the biosynthesis of natural products. Thus,

different methods for the assessment of the sustainability of bioprocesses at early stages of development,

both from an economic and environmental perspectives, as well as technologies for its intensification are

explored throughout the thesis.

In the following sections some background information on the type of processes and products under study

is given.

1.1 Background

Over the past years, there has been a generalized perception of the emergence and expansion of a so-called

bioeconomy. Government policies and industry sustainability programs are driving a bio-based economy all

over the world, supported by a growing concern about sustainability from consumers.

But what is this bioeconomy? According to European Commission, the bioeconomy encompasses the

production of renewable biological resources and their conversion into food, feed, bio-based chemical

products and bioenergy. It thus includes agriculture, forestry, fisheries and food production as well as parts

of the chemical, biotechnological and energy industries. Bioeconomy sectors are expected to have a strong

innovation potential, using a wide range of sciences (life sciences, agronomy, social sciences, etc.) and

technologies (biotechnology, nanotechnology, information and communication technologies) (European

Commission, 2014).

The EU estimates that the sectors that comprise the bioeconomy account for 22 million jobs, which is

approximately 9% of the EU’s workforce (European Commission, 2013). One of the goals of the European

commission on research, science and innovation is to move from a fossil-based economy to a bioeconomy.

And bio-based chemicals are expected to constitute the largest segment of potential growth for industrial

bio-based products. The bio-based chemical sector achieved a market value of $3.6 billion in 2011 and it is

estimated that by 2021 the global market for bio-based chemicals will have increased to $12.2 billion. While

this represents a significant increase, there is the potential to produce two-thirds of the total volume of

2

chemicals from bio-based material, which represents over $1 trillion annual global market (Golden and

Handfield, 2014). Such bio-based chemicals can be produced either by bioprocesses or by conventional

chemical processes using a bio-based feedstock. Therefore, there are good opportunities for the

development of novel bioprocesses for the production of chemicals in the coming years with an increasing

number successfully applied at an industrial scale. Different types of bioprocesses and examples of

bioprocesses that have been successfully applied in industry are given in section 1.2.

The increasing interest in bio-based products and the definition of a bioeconomy strategy by European

Commission is explained by the many challenges that Europe and the world is nowadays facing, such as an

increasing world population; depletion of many natural resources, namely fossil fuels; the impacts of

increasing environmental pressures and climate change (European Commission, 2013). But it is clear that one

driver for the expansion of bioeconomy is also the consumers demand for product transparency.

Governments are pushing for sustainability labeling of products, for instance, the Grenelle Law II in France

covers the quantification of environmental impacts and the communication of the environmental footprint

of the products to the consumers via labeling. And major multinational retailers and brands are also

undertaking life cycle assessments of their entire supply chain (Golden and Handfield, 2014). Hence,

improved environmental performance of industrial and consumer goods is a significant driver of the

development of bio-based chemicals.

However, there is also a need for more effective quantification of how the expansion of bioeconomy might

impact the environment, namely land use and water quantity and quality. A study by Weiss and co-workers

compared 44 LCA studies of bio-based materials and showed that bio-based materials saves, relative to

conventional materials, 55±34 gigajoules of primary energy and 3±1 CO2 equivalents of greenhouse gases.

However, bio-based materials may also increase eutrophication by 5±7 kilograms of phosphate equivalents

and stratospheric ozone depletion by 1.9±1.8 kg of nitrous oxide equivalents. The significant errors

associated to this values, as well as the variability in the results of LCA studies stresses difficulties in drawing

general conclusions but in any way it is important to take these environmental aspects into account when

developing processes for production of bio-based chemicals (Weiss et al., 2012). Different tools for

assessment of sustainability of bioprocesses are further described in Chapter 2.

3

1.2 Bioprocesses

Bioprocesses include both fermentation and biocatalytic processes and in some cases it might not be

straightforward to distinguish them.

In the context of this thesis biocatalytic processes are those applying enzymes as catalysts in the conversion

of one main substrate to a product in one or more steps. The biocatalyst can be used in two different forms:

i) isolated enzymes, or ii) whole-cells, which can be growing or resting. These two forms of biocatalysts can

then be used in different formulations - liquid formulations or immobilized on or into larger particles of inert

support materials.

Fermentation processes on the other hand correspond to the production of products associated with cell

growth, which are produced by primary or secondary metabolic pathways. Primary metabolic pathways

produce compounds with low molecular weight, such as alcohols, aldehydes, ketones and carboxylic acids.

Examples of products of secondary metabolic pathways are amino acids, proteins or antibiotics (Villadsen,

2015).

Examples of both types of bioprocesses are given in the following sections.

1.2.1 Biocatalytic processes

Biocatalytic processes are those that apply enzymes, which can be isolated or used as whole-cells, as catalysts

for the synthesis of chemical products.

Enzymes have an exquisite advantage over chemical catalysts, - they have a very high selectivity (enantio-,

regio- and chemoselectivity). As a result of this excellent selectivity, enzymes can be used in complex

reactions without the need of blocking and deblocking steps that are common in conventional organic

synthesis. This further results in fewer by-products and consequently reduced waste production (Schmid et

al., 2001).

Biocatalytic processes also operate at milder reactions conditions, with moderate temperatures, pH and

pressures and generally run in aqueous media, preventing large consumption of organic solvents. Unlike

many of the metal and organometallic catalysts conventionally used in chemical industry, enzymes are non-

4

toxic catalysts and are produced by fermentation from renewable feedstocks. All these features can make

biocatalysis an environmentally friendly alternative to conventional chemical catalysis.

But the most promising advantage of biocatalysis over conventional chemical catalysis is the possibility of

modifying the catalyst and improve it to have a set of desired properties. When using a biocatalyst there is

the freedom to improve the selectivity, stability and activity of the enzyme, and therefore improve reaction

rate and reaction yield. The initial phase of enzyme discovery involves screening available microbial culture

collections or libraries of enzymes from a variety of sources in order to identify an enzyme with activity

towards the substrate of interest. In some cases the wild-type enzyme may have the required activity and

selectivity to be applied in a reaction system, but the most likely scenario is that it will require some

improvements in catalytic activity, selectivity and stability under process conditions that are required for

practical applications. Directed evolution of enzymes by iterative cycles of random mutagenesis, protein

expression and screening or selection of the desired activity emerged as a powerful technology that allows

significant improvements of an existing enzyme (Turner, 2009). Enzyme improvement and optimization is

however a time consuming process and it can take between 3 to 15 months to develop a new biocatalyst

employing a team of skilled scientists (Tufvesson et al., 2010). Still, in many cases improvement of the

biocatalyst is essential to ensure an industrial application of the biocatalyst and there are several examples

where this has been successfully achieved, particularly in pharmaceutical industry (Bornscheuer et al., 2012).

Table 1.1 shows some examples of biocatalytic processes that have been successfully developed and applied

in industry.

5

Table 1.1 Some examples of industrial biocatalytic processes.

Product Enzyme Class Technology Reference

(S)-ethyl-4-chloro-3-hydroxybutyrate, an

intermediate for atorvastatin (API of the cholesterol lowering

drug Lipitor®) manufacture.

Ketoreductases (KREDs)

EC 1.1.1.2

Various protein engineering approaches, including directed evolution, to increase activity,

stability and coenzyme specificity. The volumetric

productivity per biocatalyst loading was improved 2500 fold, resulting in a 14-fold

reduction in reaction time, a 7-fold increase in substrate

loading, 25-fold reduction in enzyme use and 50%

improvement in isolated yield.

(Ma et al., 2010)

The product is sitagliptin, an antidiabetic drug.

Transaminases EC 2.6.1.X

Protein engineering to develop enzyme with initial activity,

followed by activity improvement in increasing

DMSO, substrate and isopropylamine, as well as increased thermostability.

(Savile et al., 2010)

L-fluoro-leucine, an intermediate in odanacatib

manufacture. Odanacatib is a cathepsin K inhibitor, which is

an important class of antiresorptive agents that help

prevent bone loss.

Hydrolases EC 3.X

Enzyme immobilization and reaction engineering increased

stability and activity under process conditions. The

process has been demonstrated several times at 100 kg scale (>90% yield and

88% ee).

(Truppo and Hughes, 2011)

Biocatalysis also shows an exquisite advantage over fermentation processes, which is the possibility to run a

reaction with unlimited space-time yield. Typically in aerobic fermentations, space-time yield is limited by

the oxygen consumption required for cell growth. But by disassociating the desired reaction from cell growth

it is possible to achieve a much higher space-time yield.

6

1.2.2 Fermentation processes

Fermentation processes involve de novo production of a molecule from a carbon and energy source, such as

glucose via primary metabolism, rather than conversion of a precursor molecule to the desired product by

enzymes or whole cells (Straathof et al., 2002).

Fermentation processes can be used to produce a wide range of products in all sectors of chemical industry

– bulk, fine and pharmaceutical chemicals, with varying production prices and volumes and some examples

are given in Table 1.2.

Table 1.2 Examples of industrial products produced by fermentation (adapted from (Heinzle et al., 2008)).

Product Annual production (tons/year) Price range (€/kg)

Ethanol 99,000,000 0.2

Citric acid 1,700,000 1

Proteases 100,000 3

Aspartame 10,000 5

Tetracyclines 5000 55

Insulin 8 120,000

According to their size, fermentation products can be sub-divided into small molecules and large molecules.

Small molecules have a molecular weight of 30-600 Da and a radius smaller than 1 nm. Examples of small

molecules are sugars, amino acids, organic acids or vitamins. Large molecules include proteins, nucleic acids

and polysaccharides and have a molecular weight of 103-1010 Da and a radius typically larger than 1 nm.

Among small molecules, a distinction can also be made between primary and secondary metabolites. Primary

metabolites, e.g. organic alcohols and acids are produced in the primary growth phase of the

microorganisms, while secondary metabolites are formed at or near the stationary phase. One classic

example of secondary metabolites are antibiotics. However, in many cases the distinction between primary

and secondary metabolites is not straightforward (Heinzle et al., 2008).

The microorganisms used as production hosts for fermentative production of chemicals can also be

numerous, being filamentous fungi, the yeast Saccharomyces cerevisiae or the bacteria E. coli some of the

most common ones.

7

Some fermentation processes, such as the production of citric acid or ethanol show remarkably high yields,

productivities and product concentrations and have become the conventional method for production of

these chemicals, overcoming chemical production methods. Some successful examples of fermentation

processes are shown in Table 1.3.

Table 1.3 Some examples of commodity chemicals produced by fermentation technology. The achieved yield, product concentration and space-time yield correspond to the highest values that have been

reported in scientific literature (adapted from (Straathof, 2014)).

Product Production

organism

Production by

fermentation

(million

tons/year)

Theoretical

yield (g/g)

Achieved

yield

(g/g)

Product

concentration

(g/L)

Space-time

yield

(g/(L.h))

Citric acid Aspergillus niger 1.7 1.07 0.88 240 1.4

Ethanol Saccharomyces

cerevisiae 99 0.51 0.50 170 82

Lactic

acid Lactobacillus sp. 0.37 1.00 1.00 231 150

Developing a commercially relevant fermentation process is a challenging task, especially if the product is a

bulk chemical. Space-time yield, product concentration and the yield of product on carbon source determine

the economy of these processes.

One of the main difficulties when developing a fermentation process is to be able to transfer the required

amount of oxygen for cell growth. The rate of cells growth is often limited by the rate of dissolution of oxygen

from air bubbles into the liquid and this topic has been subject of research for decades (Tsao and Lee, 1975).

The oxygen transfer limitations is translated in the differences in space-time yield between aerobic and

anaerobic fermentations. Ethanol and lactic acid production are good examples of anaerobic fermentation

processes, where the space-time yield that is achieved is extremely high (see Table 1.3). On the other hand,

citric acid is a typical example of an aerobic fermentation where space-time yield does not exceed 2 g/(L.h).

Further limitations in the development of fermentation processes are the cost and availability of feedstocks,

and a lot of research has been done in the development of fermentation processes from alternative

8

feedstocks, e.g. lignocellulosic materials; and metabolic engineering of the cells so that more products can

be produced from renewable bio-based materials by fermentation (Woodley et al., 2013).

1.3 Natural products

Natural products constitute an extensive family of diverse organic molecules with more than 200,000

compounds discovered and extracted from a variety of natural sources. (Pearsall et al., 2015).

Natural products can be sub-divided into two categories – primary and secondary metabolites. Primary

metabolites are essential for all types of living cells (plants, microbial and animal) and include sugars, fatty

acids, amino acids, nucleic acids as well as all the chemicals considered ubiquitous for plants growth and

development, such as regulators, cell wall components and metabolites unique to photosynthetic organelles

like chloroplasts. Secondary natural products are structurally and chemically much more diverse than primary

natural products, they outnumber primary natural products by orders of magnitude and are named

secondary because they have no established role in growth and development (Wu and Chappell, 2008). There

are several different classes of natural products, that can be either primary or secondary metabolites:

polyketides; non-ribosomal peptides; alkaloids; phenylpropanoids, which include flavonoids, coumarins,

lignans and stilbenes; fatty acids and terpenoids, also known as isoprenoids (Pearsall et al., 2015).

Natural products are compounds that are structurally very complex, have multiple chiral centers and can

have different biological activities. The vast number of secondary natural products is thought to be a result

of evolution, for instance, for development of plant defense responses to predators, parasites and diseases.

For example, capsidiol is an antimicrobial sesquiterpene biosynthesized by tobacco cells and tissues in

response to microbial threats (Wu and Chappell, 2008).

Natural products can have several interesting properties and they have been particularly important for

pharmaceutical industry, delivering novel antibiotics, hormones and antitumor agents to the list of

therapeutic drugs. Over the last 25 years, nearly two-thirds of all newly approved drugs (between 1981 and

2012) were derived or inspired by natural products structure (Newman et al., 2003) (Newman and Cragg,

2007).

The richness of natural products lies within the structural complexity associated with these compounds. The

possible benefits of natural products to modern society are however hindered by their relatively poor

abundance in nature, which restricts their use in research and development as extraction from natural

9

sources has very low yields. Besides the low concentration of natural products in plants, they tend to

accumulate over long growth periods, often as mixtures of hundreds of different compounds, and the

quantity and quality of natural products also varies with geographical location and climate conditions (Wu

and Chappell, 2008). All these features make the extraction of natural products economically unfeasible in

most cases.

Moreover, the structural complexity of these compounds presents challenges for chemical synthesis and

most conventional organic chemistry methods are impractical for the synthesis of natural products (Pearsall

et al., 2015). Synthetic chemists have developed means to synthesize some natural products, but often with

relatively low yields and resulting in racemic mixtures at each step in multistep synthesis. Coupling together

multiple synthetic routes of such complexity results in very low yields.

Expressing the genes responsible for the biosynthesis of these natural products and inserting them into

fermentable organisms like yeast and bacteria represent an interesting alternative to meet these production

limitations. Thus metabolic engineering is one of the areas where great success on the biosynthesis of natural

products has been achieved - namely the production of terpenoids, carotenoids, vitamin E and flavonoids

(Ajikumar et al., 2010) (Martin et al., 2003) (Anthony et al., 2009) (Pfeifer et al., 2001) (Katsuyama et al.,

2007) (Hawkins and Smolke, 2008) (Minami et al., 2008) (Lee and Schmidt-Dannert, 2003) (Das et al., 2007).

Enhancement of the production of natural products may involve the import of heterologous genes and

overexpression of required native enzymes; increasing the cell uptake of essential substrates; optimizing

pathways by minimizing alternative product formation; increasing cofactor availability and augmenting

substrate usage; optimizing enzyme activity and specificity by protein engineering or directed evolution of

enzymes and optimizing access to the final products by secretion or cell lysis methods. Examples of tools that

are typically used to tackle these aspects include overexpression of enzymes mediating rate-limiting

reactions, deregulation of gene expression, introduction of regulatory transcription factors, heterologous

pathway construction, optimizing codons for efficient protein expression, use of feedback-resistant enzymes,

exclusion of opposing pathways to minimize by-product formation and prevention of product degradation

(Yadav et al., 2012).

The three possible routes for production of natural products are represented in Figure 1.1.

10

Option 3: Biosynthesis

Product Recovery and Purification

Solid-Liquid Extraction

Biomass

Final product

Option 1: Extraction from natural sources

Chemical reaction

Reagents

Option 2: Chemical synthesis

Chemical catalyst

Fermentation

Biocatalytic reaction

Biocatalyst

Substrate

Fermentation

Carbon source

Trace metals

Oxygen

Nitrogen source

Carbon source

Trace metals

Nitrogen source

Oxygen

Solvents

Figure 1.1 Schematic representation of all the possible routes for natural products manufacturing.

11

One particular class of natural products where there has been a lot of focus on research are the isoprenoids

or terpenoids. This is the largest class of natural products with more than 40,000 different molecules

identified. Many terpenoids are commercially interesting because they can be used as flavors or fragrances

in foods and cosmetics, e.g. menthol; they can have medicinal properties, e.g. paclitaxel, an anti-cancer drug;

or have agricultural significance, e.g. cucurbitacins have shown to be involved in insect resistance (Aharoni

et al., 2005).

Two examples of successful natural products production by heterologous gene expression are i) the

production of the antimalarial drug artemisinin; and ii) the production of paclitaxel, a powerful anti-cancer

drug.

Starting with paclitaxel, also known as Taxol, this is a natural diterpenoid alkaloid that was first isolated from

the bark of the pacific yew tree Taxus brevifolia. It is a powerful anti-cancer drug that was first prescribed for

the treatment of ovarian and breast cancer, but later on was found to be effective in the treatment of other

types of cancer. It is nowadays commercialized by Bristol-Myers Squibb under the name Taxol®. As for most

natural products, extraction from the natural sources is not a viable option – the extraction yield is only 1

gram of paclitaxel per 10 kg of dry barks (Fatima et al., 2014). Taxol’s structural complexity limited its

chemical synthesis. It was first synthesized in 1994 by Nicolaou and co-workers in 51 steps but with a final

yield of 0.002% (Nicolaou et al., 1994). Later on a method for plant-cell based culture has been developed

and Taxus plant cell cultures have been scaled-up up to 7,500 liters and are currently commercially employed

by the company Phyton Biotech (Germany). However, Taxus plant cell cultures are still limited for large-scale

commercial use because of the low and unstable yield of paclitaxel, high production cost and low selectivity

over unwanted by-products. Moreover, plant cell cultures display a large degree of heterogeneity in

secondary metabolite production capabilities (Frense, 2007). Recent developments in metabolic engineering

and synthetic biology allowed the expression of the metabolic pathway for Taxol precursor, taxadiene, in

yeast (Engels et al., 2008) and in E.coli, the last with a very promising titer of approximately 1 g/L (Ajikumar

et al., 2010).

Another interesting progress made in the production of natural products is the biosynthesis of artemisinin,

an antimalarial drug. Artemisinin is naturally produced by the plant Artemisia annua, which has a long history

of use in traditional Chinese medicine. The antimalarial properties of A. annua extracts were rediscovered in

the 1970s and later on the active compound artemisinin was identified as well as its structure. Artemisinin

derivatives were recognized as anti-malarial drugs by World Health Organization in 2002, but artemisinin

extraction from the plants showed several complications, namely fluctuations in artemisinin availability and

price which is a major concern given the fact that malaria is predominantly a disease of the developing world.

12

Thus, the Semi-synthetic artemisinin project was born as a partnership between the University of California

(Berkeley, USA), Amyris Inc. and the Institute for OneWorld Health and funded by Bill and Melinda Gates

Foundation. The project has been able to successfully engineer both E. coli and yeast Saccharomyces

cerevisiae to produce artemisinic acid, which can later on be converted to artemisinin by conventional

organic chemistry conversion. The engineered yeast was selected as the production host and is able to

produce 25 g/L of artemisinic acid. Industrial production of semi-synthetic artemisinin (semi-synthetic

because the last steps are accomplished by organic chemistry) by the company Sanofi has already successfully

started (Paddon and Keasling, 2014).

The biosynthesis of the antimalarial drug artemisinin is a very successful case of heterologous gene

expression for microbial production of natural products. However, the number of natural products that are

commercially available is still very limited and more efforts are required for the development of synthetic

approaches for scaling up the production of bioactive natural products (Fatima et al., 2014).

1.4 Scope of the thesis

Even though there is an enormous potential in natural products and a big interest in their commercialization,

the task of developing an industrial process for manufacturing of natural products is very challenging. This

thesis aims at defining some tools for benchmarking of processes for natural products production, as well as

guidelines for the development of biosynthetic processes for a successful intensification and implementation

at industrial scale.

The set of tools established in this thesis can be used to evaluate processes for natural products production

in an early stage of development, both from an economic and an environmental point of view. These

evaluation tools allow benchmarking of different routes and process screening of the different alternative

processes for production of a given natural product, as well as determination of the main process bottlenecks

which can provide guidance for research and development.

This systematic evaluation approach is illustrated with two different case-studies that involve the production

of two different natural products – i) the bioconversion of (R)-limonene to (R)-perillic acid by a whole-cell

biocatalyst, and ii) the fermentation of Pichia pastoris for brazzein production.

The main components of this thesis are schematically represented in Figure 1.2.

13

Figure 1.2 Schematic representation of the scope of the thesis.

1.5 Thesis outline

The thesis is organized into three main parts, as described below.

Part I (Chapter 2) – Provides an introduction and background to the different tools for economic and

environmental assessment of bioprocesses at different stages of development.

Part II (Chapter 3-4) – Focuses on all the work developed for the two case-studies.

Chapter 3 is dedicated to case-study 1 – the bioconversion of (R)-limonene to (R)-perillic acid by a

whole-cell biocatalyst.

It includes the route selection for perillic acid production and an evaluation of the process both from

an economic and an environmental perspective. Additionally, experimental work for characterization

of the reaction system is also presented as well as a discussion of the main bottlenecks of the process

and strategies to improve it.

Chapter 4 is dedicated to case-study 2 – the fermentation of Pichia pastoris for brazzein production.

Case-studies

(R)-perillic acid

Brazzein

Economic

Process metrics

Process modelling

Environmental

Green Chemistry metrics

14

This chapter is focused on the comparison of different processes for sweeteners production from an

environmental perspective. It aims to provide guidelines for implementation of environmental

assessment of bioprocesses in early stages of development as a tool for process development.

Part III (Chapter 5-7) – Includes a general discussion of the methodology for evaluation of bioprocesses for

natural products biosynthesis in an early stage of development. A discussion of the applicability of the

different evaluation tools is included, as well as conclusions and future perspectives for this work.

15

Part I

Evaluation tools

16

17

2. Sustainability assessment of bioprocesses

Sustainability focuses on a triple bottom line - the integration of environmental integrity, social responsibility

and economic viability, represented in Figure 2.1. This triple bottom line is also known as the three Ps: Profit,

Planet and People. The three pillars of sustainability complement each other and when talking about

sustainability one must take all the three parts into consideration.

Figure 2.1 Interaction between the three parts of sustainability: economic, environmental and social (adapted from (Heinzle et al., 2008)).

The interactions between the three parts of sustainability are many and complex. For instance, the raw

material availability depends on the depletion of natural resources (environmental issue). On the other hand,

the availability of a certain raw material affects its price, which in turn affects the economy of the process.

Both economic and environmental assessments, which are further developed in the following sections, were

applied to the case-studies considered in this thesis. The social aspects of sustainability have not been

included in neither of the two case-studies.

18

2.1 Economic assessment

2.1.1 Process metrics

The economic feasibility of a biocatalytic process is of vital importance and it determines if a process can be

scaled-up to industrial manufacturing or not. The performance of a bioprocess, either a biocatalytic or a

fermentation process, can be assessed using a set of metrics that can guide biocatalyst and process

development in early stages of development (Lima-Ramos et al., 2014b) (Van Dien, 2013).

In the case of biocatalytic processes these metrics are reaction yield, biocatalyst yield, space-time yield or

volumetric productivity and product concentration. The target values for these metrics are influenced by the

market value and annual production of the product of interest, which determine the scale of the production

process. The minimum values for each process metric are case specific. They depend on the type of catalyst

used, whether it is an isolated enzyme or a whole-cell biocatalyst, the biocatalyst formulation (liquid or

immobilized) and the type of product and it is therefore difficult to define general targets for biocatalytic

processes.

Reaction yield measures the amount of product formed per amount of substrate consumed and it should be

expressed on a molar basis (% mol product/mol substrate). Reaction yield has been inaccurately defined as

a ratio between the mass of product and the mass of substrate in previous work from Lima-Ramos and co-

workers and Lundemo and co-workers (Lima-Ramos et al., 2014b) (Lundemo and Woodley, 2015). However,

reaction yield should always be defined as a molar ratio and in the particular case of natural products, it does

not make sense to calculate a mass ratio since most of the times the product has a higher molecular weight

than the substrates and therefore using a mass ratio could lead incorrectly to a reaction yield higher than

100%. Reaction yield represents a measurement of raw materials cost contribution and therefore this metric

is of greater importance for bulk chemicals, which have a low selling price and therefore the gap between

the cost of raw materials and a profitable product cost is low. The production costs are often dominated by

the cost of raw materials which puts pressure on reaction yield. The other situation where reaction yield can

be of major importance is if the substrate is a high-value chemical and the price difference between substrate

and product is low. The reaction yield is determined by the substrate concentration in the reaction media

and by the enzyme specificity. High reaction yields might be difficult to achieve in many biocatalytic reactions

due to catalyst inhibition or instability (Lima-Ramos et al., 2014b) (Ramesh et al., 2016).

19

Product concentration measures the downstream processing costs, given that for lower product

concentrations, a high volume of reaction mixture is sent for product recovery and purification. Therefore,

there will be a high cost for removing a high amount of solvent from the reaction mixture for product recovery

(Lima-Ramos et al., 2014b).

Demands on reaction yield and final product concentration are relevant in processes with isolated enzymes

or whole-cells. Space-time yield (STY) is however more important for processes with growing whole-cells in

the same way as it is for fermentation processes. Space-time yield or volumetric productivity corresponds to

the mass of product formed per reaction volume per reaction time and is typically expressed in g/L/h. It

accounts for capital costs, namely for the reactor cost, and energy and labor related costs during the reaction

(Lundemo and Woodley, 2015).

For resting cell processes or processes with purified and isolated enzymes, biocatalyst yield is applied instead.

This metric represents the total amount of product that can be produced per mass of biocatalyst, similarly to

the total turnover number. It reflects the cost of biocatalyst production (fermentation and biocatalyst

purification and formulation) on the biocatalytic process. Hence, its values are influenced by the final cell

density and protein expression of the fermentation process for enzyme production. Catalyst recycle is one of

the possible strategies to improve the biocatalyst yield (Lima-Ramos et al., 2014b) (Ramesh et al., 2016)

(Lundemo and Woodley, 2015).

Several successfully implemented industrial biocatalytic processes show product concentrations higher than

50 g/L or space-time yields higher than 1 g/L/h. Straathof and co-workers reviewed 134 biotransformation

processes for production of fine chemicals, both with isolated enzymes or whole-cells as catalysts, and

reported space-time yields well over 1 g/L/h, reaction yields typically well over 80% and mostly over 90%

(exceptions are kinetic resolutions which give yields close to 50%) and final product concentrations well

above 50 g/L. Product concentrations varied from 50 g/L to above 200 g/L for carbohydrates and amides. The

differences in product concentrations correlated with product solubility, toxicity and/or stability (reaction

systems where product inhibition was seen showed a lower product concentration) (Straathof et al., 2002).

The biocatalyst yield is dependent on the enzyme formulation. For immobilized enzymes or immobilized

resting cells, yields of 102-104 kg per kg of catalyst have been reported (Thomas et al., 2002).

In order to achieve process metrics that allow an economically feasible implementation of the process, it is

clear that in most cases a strategy for biocatalyst improvement (via targeted protein engineering) should be

complemented by the implementation of process intensification strategies into the reaction and process of

interest (Ramesh et al., 2016). The development of a biocatalytic process should then be done in two main

20

steps - first by process intensification followed by process scale-up and increase of the production volume.

In the process intensification stage, the process is kept at a small scale but the process metrics are improved

by enzyme modification and reaction and process engineering tools. Once the process metrics have achieved

the desired values, there is confidence that the process is indeed scalable and the volume of production can

be scaled-up (Lima-Ramos et al., 2014a).

Regarding fermentation processes, product concentration and space-time yield are also critical metrics to

evaluate them and are in the same range of values as for biocatalytic processes with growing cells

(Oudshoorn et al., 2010). The third metric that is used to assess fermentation processes is the yield of product

on the carbon source, measured in %(C-mol/C-mol) or %(g/g). In this case, the rule of thumb is that the yield

needs to be higher than 80% of the maximum theoretical yield (calculated from the metabolic pathway) for

bulk chemicals (Van Dien, 2013).

Several examples of microbial cells whose metabolism is engineered to produce chemicals and/or fuels, as

well as novel enzyme-based reaction systems are described in scientific literature. However, it is still a

relatively small number of them that reaches commercialization. At early stages of research, cost calculations

and understanding the influence of process metrics on the production cost are critical to estimate targets to

obtain an economically feasible process with a given production organism (van’t Riet, 1986). It is difficult to

estimate the cost and research intensity necessary to modify the production microorganism or enzyme. But

having that, one can weight it against the targets for the key process metrics and estimate the cost and

development time necessaries to achieve industrial implementation.

2.1.2 Full cost estimation

Economic evaluation is a tool that is often used in chemical engineering to quantitatively estimate the costs

and the expected profitability of a given chemical process. It includes cost estimation, revenue estimation

and profitability analysis.

Typically cost estimation is made throughout the development process of a chemical process and the

accuracy of the estimations increase in later development stages. Cost estimation in a preliminary design

stage involves the application of rules-of-thumb and rough estimations (e.g. for equipment cost), which

means that the accuracy of estimations at this stage of development is somehow dependent on the

experience of the engineer working on this task. Typically in early-stages of development the accuracy is

about ±20-30%. As the project design is refined, the estimates become more detailed and accurate. But even

21

with a certain degree of uncertainty it is extremely useful to estimate the costs of a chemical process during

its development stages since it facilitates process debottlenecking and can guide research (Lima-Ramos et

al., 2014b).

Cost estimation is divided into two categories: estimation of capital expenditure or capital investment

(CAPEX) and operating expenditure or operations cost (OPEX).

CAPEX corresponds to the costs of getting the industrial plant into operation. It includes the fixed-capital

investment (FCI), which refers to the capital necessary for all the equipment and accessories required in the

process operation and start-up; and working capital, which corresponds to the sum required for day-to-day

operation of the plant, including the cost of inventories and supplies.

Operation costs (OPEX) consist of variable production costs; fixed charges and plant overhead costs. Variable

production costs include raw materials costs; labor costs; utilities; patents and royalties; waste treatment

and disposal; laboratory, quality control and quality assurance. Fixed charges or fixed costs include

depreciation of equipment and buildings, local taxes, property insurances and rent and interests. Finally,

plant overhead costs include general expenses, such as administrative costs, distribution and marketing, and

research and development (R&D) costs (Peters et al., 2004). Figure 2.2 includes a scheme of all categories

and sub-categories of cost estimation.

Since in early stages of development the level of uncertainty is high, it is also necessary to evaluate the effect

of certain modifications to the original process on the total process environmental and economic profile by

performing a sensitivity and/or uncertainty analysis (Lima-Ramos et al., 2014b). Developing a process model

even in early stages of development can be extremely useful, since it can be a tool to evaluate the influence

of certain process parameters on the final production cost (OPEX).

There are several different softwares for process design and simulation that facilitate cost estimation, since

they incorporate an estimator of equipment prices. Examples of these softwares are Aspen Plus®, from Aspen

Technology, Inc.; SuperPro Designer®, from Intelligen Inc.; or SimSci PRO/II from Schneider Electric. In this

thesis SuperPro Designer® (version 9.5) was selected as the software for process design, mass and energy

balances and cost estimation.

22

Figure 2.2 Cost estimation different categories and sub-categories.

23

2.2 Environmental assessment

Over the past decades there has been a rise in public awareness of health and environmental issues of

consumer goods. Most companies recognize this trend and feel continued pressure to find greener

alternatives. Policy makers are also increasingly concerned with sustainability, which is translated in one hand

on tighter regulations on the use of certain raw materials and on the emissions resulting from chemical

production processes, and on the other hand in an increase of interest for research projects aiming to

develop sustainable and “green” chemicals. This trend has led many companies to evaluate the sustainability

of their products and production processes and to include environmental concerns as a driver for product

and process development.

Both companies and academia are today faced with the challenge of having to evaluate the environmental

impact of a certain chemical synthetic route and/or production process. Hence, there is the need for

developing methods for assessing the environmental profile of chemical production processes in different

development stages, which is a rather complex task. Different tools for the environmental assessment of

chemicals processes, with varying degrees of detail and complexity have been developed and are nowadays

applied by many chemical companies, in particular pharmaceutical companies, to assess their processes. On

one hand there are reaction- and process-related metrics, which try to comparatively assess processes from

a green chemistry and green engineering perspectives. On the other end of complexity we have the Life Cycle

Assessment (LCA) methodology – see Figure 2.3.

24

Figure 2.3 Correlation between process development stage and the complexity of the environmental assessment.

Given the complexity of these different tools and the amount and type of data required for each of them,

they should be applied in different stages of development of the chemical process.

Green Chemistry metrics try to quantify and measure the “greenness” of a given process or reaction by

translating it in terms of simple metrics which should be easy to understand and apply. The amount of data

required is relatively low, which makes it an appropriate tool for synthetic route selection and early stages

of research and process development. These simple metrics lack however a holistic view of the

environmental impacts of the product in its whole life cycle and do not distinguish the different types of

waste and emissions generated.

LCA on the other hand is a very detailed tool for evaluation of the product environmental burden in its entire

life cycle (including raw materials manufacturing, manufacturing process of the desired product, use,

transportation and final disposal). It is a very work intensive, complex and time-consuming tool and is

therefore adequate for later stages of development, such as the process implementation and marketing.

Some methodologies aiming to guide the environmental assessment of chemical processes in its different

development stages have been published. Examples of these are the metrics toolkit resulting from the

CHEM21 project for evaluating the “green credentials” of a certain reaction, developed by McElroy and co-

workers (McElroy et al., 2015), or the methodology to guide the development of biocatalytic processes

25

developed by Lima-Ramos and co-workers (Lima-Ramos et al., 2014b). There are slight differences in the

methodologies but it is widely accepted that the level of detail of the environmental assessment should

increase with the level of development of the chemical process.

The different tools for environmental assessment are described in more detail in the following sections.

2.2.1 Green Chemistry

Green chemistry is a concept and guiding principle which aims to encourage the development of

manufacturing processes and products with the lowest possible environmental impact. The twelve principles

of green chemistry were introduced in 1998 by Anastas and Warner and are summarized in Table 2.1 (Anastas

and Eghbali, 2010).

26

Table 2.1 The twelve principles of green chemistry (adapted from (Anastas and Eghbali, 2010)).

Prevention It is preferable to prevent waste than to treat or clean up waste after it is formed.

Atom economy Synthetic methods should be designed to maximize the incorporation of all materials used in the process into the final product.

Less hazardous chemical synthesis Whenever possible, synthetic methods should be designed to use and generate substances that pose little or no toxicity to human health and the environment.

Designing safer chemicals Chemical products should be designed to preserve efficacy of the function while reducing toxicity.

Safer solvents and auxiliaries The use of auxiliary substances (e.g. solvents) should be avoided whenever possible and, when used, innocuous.

Design for energy efficiency

Energy requirements of chemical processes should be recognized for their environmental and economic impacts and should be minimized. If possible, synthetic methods should be conducted at ambient temperature and pressure.

Use of renewable feedstocks A raw material or feedstock should be renewable rather than depleting whenever technically and economically feasible.

Reduce derivatives

Unnecessary derivatization (use of blocking groups, protection/deprotection, temporary modification of physical/chemical processes) should be minimized or avoided if possible, since such steps require additional reagents and can generate waste.

Catalysis Catalytic reagents (as selective as possible) are superior to stoichiometric reagents.

Design for degradation Chemical products should be designed so that at the end of their function they break down into innocuous degradation products and do not persist in the environment.

Real-time analysis for pollution prevention

Analytical methodologies need to be further developed to allow for real-time, in-process monitoring and control prior to the formation of hazardous substances.

Inherently Safer Chemistry for accident prevention

Substances and the form of a substance used in a chemical process should be chosen to minimize the potential for chemical accidents, including releases, explosions and fires.

These principles go hand in hand with the economy of a process, since the more efficient a process is, the

lower waste production is. Figure 2.4 relates waste production, expressed as the difference between the

27

total mass of materials used in the production process (M) and the mass of final product (P), with process

efficiency, which is expressed as the ratio P/M. An ideal process (which is thermodynamically impossible)

would have “zero waste production”, i.e. all the material used in the process would be converted into the

final product (P/M=1 and M-P=0). A real process can improve its efficiency and therefore reduce waste

production by means of technological improvements to the process and/or change of the raw material used

(Lapkin and Constable, 2009).

Figure 2.4 Relation between waste produced and process efficiency (adapted from (Lapkin and Constable, 2009)). M represents the total mass of materials used in the production and P the mass of final product.

2.2.1.1 The natural fit of biocatalysis with green chemistry

Biocatalysis fits most of the goals and principles of green chemistry. The high selectivity of enzymes and their

ability to carry out difficult chemical reactions eliminate synthesis steps, the associated waste in multistep

reactions and the need for protective groups and for resolutions involving chemical derivatives. It also allows

most of the starting material atoms to be converted to the product, contributing in this way to a maximization

of atom economy.

The preferential solvent used in biocatalytic reactions is water, which eliminates organic solvent waste. The

reactions typically run at neutral pH and room temperature, which reduces energy consumption. Mild

reaction conditions also minimize the potential for explosions, fires and accidental spills. Furthermore,

Was

te p

rod

uce

d (

M-P

)

Efficiency of the process (M/P)0 10

M

28

enzymes are biodegradable catalysts and are manufactured using renewable feedstocks (Tao and Kazlauskas,

2011).

An evidence that biocatalysis can contribute to green chemistry is the fact that many of the recent winners

of the US Environmental Protection Agency’s Presidential Award in Green Chemistry have used biocatalysis

as the key improvement technique. For instance, in 2012 Codexis, Inc. and Professor Yi Tang from the

University of California, Los Angeles received the award for developing a biocatalytic process to manufacture

Simvastatin, a drug for treating high cholesterol, (US Environmental Protection Agency) (Xie et al., 2006).

The “green label” given to biocatalysis is often used as a “selling argument”. However as appropriately argued

by Ni and co-workers, biocatalysis is not green per se and there is a big gap between the statement of such

green claims and the environmental profiles of reaction systems reported in scientific literature (Ni et al.,

2014). In many cases, reactions that are reported in scientific literature take place in very diluted solutions,

conditions which are not consistent with an environmentally friendly process. In these conditions large

amounts of wastewater are generated, whose treatment consumes additional reagents and energy and can

have a negative contribution to the environment (de María and Hollmann, 2015). Moreover, they often face

problems of low productivities or space-time yield, which are caused by problems of substrate and/or

product inhibition, reactant availability, biocatalyst activity and stability, etc. (Wenda et al., 2011). Claiming

that such reactions are “green” just because they fulfill one of the twelve principles of green chemistry (in

most cases atom economy) is not truthful since the 12 principles of green chemistry are intended to be used

as a cohesive system (Ni et al., 2014).

These reactions that are reported in scientific literature are still under early stages of development and

consequently in most of the times they show very high E-factors. In fact, a lot of times these biocatalytic

reactions show an E-factor higher than their conventional chemical alternative, as opposed to what would

be expected by many researchers. For instance, Kuhn and co-workers compared the biocatalytic production

of (S)-styrene with three different options of chemical catalysis and it was shown that the biocatalytic option

showed an E-factor of 3.42 kg/kg which is lower than two of the chemical options (i) ferric phenanthroline as

catalyst with 6.58 kg/kg and ii) manganese sulfate as catalyst with 5.72 kg/kg), but higher than the chemical

option using titanium silicate as the catalyst (E-factor of 1.71 kg/kg). This option was also the one with the

lowest PMI (3.04 kg/kg) as opposed to the biocatalytic option with a PMI of 21.07 kg/kg (Kuhn et al., 2010).

But in reality, enzymes did not evolve to be an environmentally friendly catalyst. Enzymes have evolved to

catalyze a vast number of reactions that contribute to the survival and reproduction of the organisms that

produce them. The fact that they can be used for our benefit in industrial production of several types of

chemicals is remarkable, but that is not enzymes primary purpose. In fact, many of the characteristics they

29

have developed during their evolution are not desirable from an industrial perspective. For instance, the fact

that many enzymes face product inhibition at certain concentrations of product is what allows them to

regulate the flow of metabolites inside the cells, which is fundamental for the cells metabolism. However,

for industrial applications this feature is unwelcome since the industrial goal is to produce high

concentrations of product in a short period of time (Arnold, 2001). Researchers force enzymes to work in

conditions that are completely different from the conditions they have evolved to work on. In nature

enzymes have evolved to work in dilute solutions, with low substrate and product concentrations, so it is not

surprising that so many enzyme-based reactions show high E-factors.

It is extraordinary to see so many novel enzyme-based reactions being reported and even though most of

them do not show an environmentally friendly profile, experience has shown that if the process is successfully

intensified by means of biocatalyst modification combined with reaction and process engineering, it is

possible to attain enzymes full potential and develop an industrial chemical process with a “green” profile.

Several examples have been reported in literature of such processes, and are shown in Table 2.2.

The biocatalytic synthesis of (S)-ethyl-4-chloro-3-hydroxybutyrate, an intermediate for atorvastatin

production, which was previously mentioned in Chapter 1, shows a total E-factor of 5.8 kg waste/kg product.

If water is included E-factor increases to 18 kg/kg, which is still a remarkably low E-factor (Ma et al., 2010).

The biocatalytic synthesis of pregabalin also shows a very low E-factor and PMI. Pregabalin is a lipophilic γ-

aminobutyric acid analogue that was developed for the treatment of several nervous system disorders, such

as epilepsy or anxiety, and is commercialized under the name Lyrica® by Pfizer. Pregabalin was first

synthesized with chemical catalysts, which produced a racemic mixture of pregabalin. The right enantiomer

was obtained by classical resolution with a very low efficiency of 30%. The E-factor of this process was

therefore relatively high – 86 kg/kg. The chemical synthetic route was then substituted by a biocatalytic route

with two lipases. The biocatalytic route reduced the E-factor to 17 kg/kg and the process was still further

improved by recycling the wrong enantiomer, (R)-cyanodiester, which further reduced E-factor to 8 kg/kg

(Dunn et al., 2010) (Dunn, 2012).

30

Table 2.2 E-factors for some examples of biocatalytic reactions at industrial scale. The E-factors shown only refer to the reaction (DSP is not included) and do not include water.

Product E-factor (kgwaste/kgproduct) Reference

(S)-ethyl-4-chloro-3-hydroxybutyrate 5.8 (Ma et al., 2010)

(S)-cyanodiester (Pregabalin) 8 (Dunn, 2012)

(R)-tetrahydrothiophene-3-ol 2.5 (Liang et al., 2010)

A quantitative evaluation of these green chemistry metrics at early stages of research and comparison with

similar processes at industrial scale can give some valuable insight to the reaction development and provide

guidance to researchers working on these reactions in the laboratory. The scientific community working on

biocatalysis can greatly benefit from knowledge sharing and reporting of more quantitative studies of the

environmental profile of biocatalytic reactions in early stages of research.

2.2.1.2 Green Chemistry metrics

The first green chemistry metrics to be introduced were atom economy by Trost in 1991 (Trost, 1991) and E-

factor by Sheldon in 1992 (Sheldon, 1992). Ever since the 90s green chemistry metrics have proliferated in

number, several new metrics were introduced (Curzons et al., 2001) (Hudlicky et al., 1999) and a considerable

amount of reviews of various metric systems and methodologies on how to assess chemical processes from

a green chemistry perspective have been published (Constable et al., 2002), (McElroy et al., 2015), (Jimenez-

Gonzalez et al., 2011), (Shonnard et al., 2003), (Roschangar et al., 2015), (Lima-Ramos et al., 2014b).

However, the green chemistry community has not benefited from the introduction of so many new metrics.

In contract, there is now an excess of metrics without standardized definitions or boundaries, which inhibits

its wide integration in industrial process research and development. There is a lack of unified metrics that

researchers both from academia and industry can easily apply in early stages of research, which is a barrier

for the fulfillment of green chemistry’s full potential (Roschangar et al., 2015).

It is however, a very complex task to select a set of key metrics that have the greatest relevance to evaluate

the “greenness” of a given process. It is difficult to find just a few metrics that are able to describe the

complexity of a given chemical/biochemical process.

31

Environmental metrics must be adapted for their context and continuously evaluated as to their utility,

applicability and relevance/appropriateness to the process under study. They should be easily understood

and accepted by key stakeholders (scientists, engineers, managers, etc.). Overall they should promote

strategic analysis and continuous improvement of a process. They should be routinely questioned, assessed,

evaluated and evolved to help make strategic decisions and make them useful to a business (Lapkin and

Constable, 2009).

Green metrics can be divided into two main groups: i) metrics that are based on the stoichiometric equation

of the chemical reaction under study and evaluate its potential environmental impact and ii) metrics that

address the amount of waste generated in the entire process (upstream and downstream processing

included).

Reaction-related metrics

Atom Economy

The first group of metrics includes atom economy (AE), which was introduced by Trost in 1991 (Trost, 1991).

Atom economy (Equation 2.1) is based on a theoretical maximum yield of 100% and it assumes that reactants

are used in stoichiometric amounts, disregarding also the use of solvents, acids or bases. It measures how

much of the starting material ends in the desired product, assessing in that way the stoichiometry of the

reaction. The driver behind this metric is to ensure that chemists design and select a production route where

as many as possible of the atoms of the substrate are included in the final product, i.e. a synthetic route with

a high mass yield and low waste (Lima-Ramos et al., 2014b).

Equation 2.1

𝐴𝐸 (%) =𝑀𝑜𝑙𝑒𝑐𝑢𝑙𝑎𝑟 𝑤𝑒𝑖𝑔ℎ𝑡 𝑜𝑓 𝑝𝑟𝑜𝑑𝑢𝑐𝑡

∑ 𝑚𝑜𝑙𝑒𝑐𝑢𝑙𝑎𝑟 𝑤𝑒𝑖𝑔ℎ𝑡 𝑜𝑓 𝑟𝑒𝑎𝑐𝑡𝑎𝑛𝑡𝑠 × 100

32

Carbon Efficiency

Carbon efficiency (CE) (Equation 2.2) was first developed by GlaxoSmithKline (GSK). It is merely a refinement

of AE that takes into account the yield and the amount of carbon in the reactants that is incorporated into

the final product. It is defined as the percentage of carbon in the reactants that remain in the final product

(Constable et al., 2002). A high carbon efficiency is desired, given that in a reaction system with a low CE

there is a greater amount of carbon that will eventually end in waste streams or as emissions (Lima-Ramos

et al., 2014b).

Equation 2.2

𝐶𝐸 (%) =𝑀𝑎𝑠𝑠 𝑜𝑓 𝑐𝑎𝑟𝑏𝑜𝑛 𝑖𝑛 𝑡ℎ𝑒 𝑝𝑟𝑜𝑑𝑢𝑐𝑡

∑ 𝑚𝑎𝑠𝑠 𝑐𝑎𝑟𝑏𝑜𝑛 𝑝𝑟𝑒𝑠𝑒𝑛𝑡 𝑖𝑛 𝑟𝑒𝑎𝑐𝑡𝑎𝑛𝑡𝑠 × 100

Reaction Mass Efficiency

Reaction mass efficiency (RME) is a metric that was also developed by GSK. RME (Equation 2.3) takes into

account both the concepts of atom efficiency, reaction yields and the molar quantities of reactants. It

corresponds to the percentage of the mass of reactants that remain in the product. RME combines key

elements of chemistry and process and represents a simple, objective, easily derived and understood metric

for use by chemists, process chemists or chemical engineers (Constable et al., 2002).

Equation 2.3

𝑅𝑀𝐸(%) =𝑀𝑎𝑠𝑠 𝑜𝑓 𝑓𝑖𝑛𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡

∑ 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑟𝑒𝑎𝑐𝑡𝑎𝑛𝑡𝑠 × 100

Both atom economy and reaction mass efficiency can be very useful for comparison of different routes for

production of a given chemical and complement each other well. Atom economy is a metric that is very

simple to calculate and one only needs to know the stoichiometric equation to calculate this metric. It tells

33

the user if the synthetic route is designed in a way that maximizes the incorporation of all the starting

materials used in the reaction into the final product. If so, atom economy should be high (ideally 100%) and

the synthetic route has a good potential to be a sustainable synthetic route. Since most of the starting

material is incorporated into the final product, the waste and emissions that are generated from this reaction

should be low.

However, this metric does not take into account the reaction yield and it disregards the use of auxiliary

substance (solvents, acids, bases etc.). Therefore, it can be misleading to look only to atom economy and this

is often the case in biocatalytic reactions. Many of these reactions show a very high atom economy and in

theory these synthetic routes have a very good potential to be “greener” when compared to conventional

chemical synthesis. However, it is also often the case that these reactions have a low reaction yield and run

in very diluted aqueous solutions. It is therefore important to complement atom economy with RME, since

this metric takes into account the reaction yield and the “real” amounts of reactants and auxiliary chemicals

used in the reaction, hence giving a more realistic picture of the environmental profile of the reaction. The

biggest disadvantage of RME is the fact that it requires experimental work to be performed prior to its

calculation.

Process-related metrics

E-factor

On the group of metrics that assess the environmental impact of the overall process, E-factor is a widely

accepted and useful metric, which was first proposed by Roger Sheldon. E-factor (Equation 2.4) is the ratio

of waste produced per unit of final product, considering that waste includes all the reactants, solvents, etc.,

basically all the chemicals used in the process, except the desired product. The higher E-factor is, the higher

waste production is and consequently, the environmental impact is greater. Ideally, E-factor should be zero.

E-factor has been widely adopted in chemical industry, particularly in pharmaceutical industry, being widely

used by companies such as GlaxoSmithKline, Pfizer or AstraZeneca.

34

Equation 2.4

𝐸 𝑓𝑎𝑐𝑡𝑜𝑟 =𝑡𝑜𝑡𝑎𝑙 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑤𝑎𝑠𝑡𝑒

𝑚𝑎𝑠𝑠 𝑜𝑓 𝑓𝑖𝑛𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡

According to Sheldon definition of E-factor, water is excluded from its calculations since in many cases its

inclusion can lead to exceptionally high E-factors that can hinder a meaningful comparison of chemical

processes. In many instances, water by itself does not constitute a significant environmental impact.

However, in certain industries, such as pharmaceutical industry, the use of highly purified water and the use

of high amounts of water in fermentation and biocatalytic processes, can result in a considerable

environmental impact. Therefore, there is a trend in the pharmaceutical industry towards the inclusion of

water in E-factor. Also because water usage becomes increasingly problematic due to shortage of potable

water in certain parts of the world (Lapkin and Constable, 2009).

In Table 2.3 typical values for E-factor for the different sectors of chemical industry are shown. Waste

production is typically greater in pharmaceutical industry, since the products are complex molecules whose

production involves multistep synthesis (Roschangar et al., 2015).

Table 2.3 E-factor (excluding water) for different sectors of chemical industry (adapted from (Tao and Kazlauskas, 2011)).

Industry sector Annual production (ton/year) E factor (kg waste/kg product)

Bulk chemicals 104-106 < 1-5

Fine chemicals 102-104 5 to >50

Pharmaceutical chemicals 10-103 25 to >100

E-factor has demonstrated a strong positive correlation to process economics. A lower E-factor has been

shown to be indicative of reduced manufacturing costs, reduced costs from hazardous and toxic waste

disposal, improved manufacturing capacity utilization and reduced energy demand in the pharmaceutical

industry, which shows how industry has strong economic incentives to integrate green chemistry into process

research, development and manufacturing (Leahy et al., 2013) (Roschangar et al., 2015).

35

Process Mass Intensity (PMI)

Process mass intensity (PMI) (Equation 2.5) takes into account the reaction yield, stoichiometry, solvents,

reagents, etc., and expresses this on a weight/weight basis. It includes all the materials used in the process

or in a specific process step. Water may or may not be included in the calculation of PMI. Similarly as to E-

factor, the trend in the pharmaceutical industry is to include it (Constable et al., 2002).

Equation 2.5

𝑃𝑀𝐼 =∑ 𝑚𝑎𝑠𝑠 𝑖𝑛𝑝𝑢𝑡 𝑚𝑎𝑡𝑒𝑟𝑖𝑎𝑙𝑠

𝑚𝑎𝑠𝑠 𝑜𝑓 𝑓𝑖𝑛𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡

The American Chemical Society Green Chemistry Institute’s Pharmaceutical Roundtable (ACS GCIPR) has

adopted PMI as the key metric for evaluating and benchmarking processes and uses it to drive more efficiency

and innovation in the pharmaceutical and fine chemical industries (Jimenez-Gonzalez et al., 2011).

PMI and E-factor are similar metrics and one can argue that they give the same type of information. However,

while in a broad sense one can argue that it does not matter which of these metrics is used, Jimenez-Gonzalez

and co-workers argues that increasing efficiency, translated in PMI, is more likely to capture management

attention than waste reduction, translated by E-factor and therefore, PMI should be preferred over E-factor.

In the business context, efficiency metrics, e.g. PMI, have an advantage over waste metrics, such as E-factor,

since they communicate and frame sustainability in terms of adding value, by adding productivity, instead of

managing costs, by waste reduction. Metrics such as PMI allow chemists and engineers to focus on how to

make a process more efficient instead of solving the problem of waste management. Focusing on waste

reduction can help chemical companies to reduce costs, whereas focusing on efficiency enables innovation

to create additional value. If the focus is on maximization of value and efficiency, waste reduction will also

consequently be one of the benefits. Hence, PMI can promote a philosophy of efficiency and innovation

instead of an end-of-pipe focus on waste (Jimenez-Gonzalez et al., 2011).

When calculating PMI one can also breakdown total PMI in its different contributors and analyze which one

of the different materials categories (solvents, reactants, water, etc.) contributes the most to total PMI and

this gives an indication of which process unit of operation requires further development and optimization.

36

It has been shown by the ACS GCIPR that PMI is improved over the course of a pharmaceutical product

development. At the transition to an advancing development phase, a decrease in PMI is observed. From the

pre-clinical studies until commercial production of an API, PMI is decreased from values in the range of 1405

kg materials/kg API to 168 kg materials/kg API (Roschangar et al., 2015).

Solvent Intensity (SI)

From the assessment of several synthetic routes of different chemicals, solvents have been found to

commonly be one of the biggest contributors to mass intensity. This is particularly relevant in the

pharmaceutical sector, where Constable and co-workers reported that solvents constitute 56% of total

materials mass used to manufacture an API and 80-90% of the total non-aqueous mass of materials

(Constable et al., 2007).

Therefore, solvent intensity (SI) (Equation 2.6) was defined as a metric to analyze and quantify the amount

of solvents used in a given process.

Equation 2.6

𝑆𝐼 =∑ 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑠𝑜𝑙𝑣𝑒𝑛𝑡𝑠

𝑚𝑎𝑠𝑠 𝑜𝑓 𝑓𝑖𝑛𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡

Since solvents are one of the biggest mass contributors to a chemical process, limiting the amount of solvents

used and selecting the “greenest” ones can result in a significant reduction of the environmental impact of a

pharmaceutical product. The selection of the greenest solvent is a compromise between different

constraints. On one hand, safety, health and environmental criteria should be considered, but also industrial

constraints (e.g. boiling point, freezing temperature, density, quality) need to be met (Prat et al., 2014).

Solvents are not directly responsible for the composition of a reaction product, nor are the active component

of a formulation. Therefore, the use of toxic, flammable and environmentally damaging solvents would seem

unnecessary because these characteristics have no impact on the function or progress of the system in which

the solvent is applied. However, these non-benign characteristics are often linked to the beneficial attributes

of the solvent needed for the application. For instance, the volatility of the solvents permits simple recovery

37

and purification of the solvent by distillation, but it also creates unwanted air emissions and the risk of worker

exposure. Amide solvents have the high polarity required to dissolve a broad range of substrates and

accelerate reactions, but this attribute often implies reproductive toxicity. Hydrocarbon solvents provide the

ability to dissolve oils in extractions and perform separations, but at the same time they are highly

combustible and their low water solubility is related to bioaccumulation and water toxicity. In attempts to

eliminate undesirable solvents, replacement strategies often seek structurally related compounds not yet

covered by the legislative and regulatory measures that are usually required to force solvents substitution.

But even though, solvents that are similar structurally can be an easy and obvious first choice for

replacement, they are likely to present many of the same environmental, health and safety (EHS) problems.

One example of these substitutions is benzene and toluene. Benzene was recognized as a carcinogen

compound in the mid-twentieth century and has generally been replaced by toluene. However, toluene is

now suspected of damaging the unborn child and of organ damage through prolonged exposure and is

included in the list of “substances of very high concern” by European Union and is likely to be banned by

REACH regulation in the future (Byrne et al., 2016).

In an attempt to categorize solvents with respect to their EHS profile, several solvent selection guides have

been published - (Tobiszewski et al., 2015), (Byrne et al., 2016), (Prat et al., 2016), (Prat et al., 2014),

(Henderson et al., 2011), including the solvent selection guides from several pharmaceutical companies, such

as GSK’s (Henderson et al., 2011), Pfizer (Alfonsi et al., 2008) and Sanofi (Prat et al., 2013). These intend to

promote the substitution of conventional organic solvents with greener solvents by academics and

researchers giving more information on solvent selection than the regulatory assessments do.

Water Intensity (WI)

A particular case of solvent intensity is water intensity (WI) (Equation 2.7), which aims to analyze the amount

of water used in the process. WI is particularly relevant in biocatalytic and fermentation processes, since it is

one of the biggest mass contributors to these processes. Even though water is an environmentally benign

solvent, it is becoming a scarce and overexploited natural resource. Furthermore, bioprocesses generate a

large amount of wastewater that needs to be treated, which is usually very energy intensive. Therefore,

water usage should also be optimized and WI appears as a metric that intends to promote a critical thinking

on water usage (Lima-Ramos et al., 2014b).

38

Equation 2.7

𝑊𝐼 =∑ 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑤𝑎𝑡𝑒𝑟

𝑚𝑎𝑠𝑠 𝑜𝑓 𝑓𝑖𝑛𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡

There are several other metrics that have been proposed to evaluate the environmental profile of a chemical

process, such as Effective Mass Yield (EMY).

Effective Mass Yield (EMY) is a metric proposed by Hudlicky and co-workers that tries to highlight the

environmental effect of the reagents. It is defined as the percentage of the mass of product relative to the

mass of all non-benign materials used in its synthesis (Equation 2.8). Hudlicky defines benign as “those by-

products, reagents or solvents that have no environmental risk associated with them” (e.g. water, dilute

ethanol, etc.) (Hudlicky et al., 1999).

Equation 2.8

𝐸𝑀𝑌(%) =𝑚𝑎𝑠𝑠 𝑜𝑓 𝑓𝑖𝑛𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡

∑ 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑛𝑜𝑛 − 𝑏𝑒𝑛𝑖𝑔𝑛 𝑟𝑒𝑎𝑔𝑒𝑛𝑡𝑠 × 100

However, this definition lacks clarity and objectivity. The applicability of this metric is also limited since there

is no information on human toxicity and ecotoxicity for all the chemicals. Besides, depending on the situation

even water, ethanol and other “benign chemicals” can have a significant environmental impact (Constable

et al., 2002).

Table 2.4 shows a summary of the most commonly used green chemistry metrics for assessment of chemical

and/or biocatalytic processes.

39

Table 2.4 Summary of green chemistry metrics (adapted from (Roschangar et al., 2015)).

Green chemistry

metric Abbreviation Equation Units

Consideration for Optimum

value

First introduced

by Reaction

yield Stoichiometry Solvents Water

Reaction-related metrics

Atom economy

AE

𝑀𝑊 𝑜𝑓 𝑝𝑟𝑜𝑑𝑢𝑐𝑡

∑ 𝑀𝑊 𝑜𝑓 𝑟𝑒𝑎𝑐𝑡𝑎𝑛𝑡𝑠 × 100 % X X X X 100%

(Trost, 1991)

Carbon efficiency

CE

𝑚(𝑐𝑎𝑟𝑏𝑜𝑛 𝑖𝑛 𝑡ℎ𝑒 𝑝𝑟𝑜𝑑𝑢𝑐𝑡)

∑ 𝑚 (𝑐𝑎𝑟𝑏𝑜𝑛 𝑖𝑛 𝑟𝑒𝑎𝑐𝑡𝑎𝑛𝑡𝑠)× 100 % X X 100%

(Curzons et al., 2001)

Reaction mass efficiency

RME

𝑚𝑎𝑠𝑠 𝑜𝑓 𝑓𝑖𝑛𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡

∑ 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑟𝑒𝑎𝑐𝑡𝑎𝑛𝑡𝑠 × 100 % X X 100%

(Curzons et al., 2001)

Process-related metrics

Environmental impact factor

E-factor

𝑡𝑜𝑡𝑎𝑙 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑤𝑎𝑠𝑡𝑒

𝑚𝑎𝑠𝑠 𝑜𝑓 𝑓𝑖𝑛𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡 kg/kg X 0

(Sheldon, 1992)

Process mass intensity

PMI

∑ 𝑚𝑎𝑠𝑠 𝑖𝑛𝑝𝑢𝑡 𝑚𝑎𝑡𝑒𝑟𝑖𝑎𝑙𝑠

𝑚𝑎𝑠𝑠 𝑜𝑓 𝑓𝑖𝑛𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡 kg/kg 1

ACS GCIPR (2007)

Solvent Intensity

SI

∑ 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑠𝑜𝑙𝑣𝑒𝑛𝑡𝑠

𝑚𝑎𝑠𝑠 𝑜𝑓 𝑓𝑖𝑛𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡 kg/kg X X X 0

(Curzons et al., 2001)

Effective mass yield

EMY

𝑚𝑎𝑠𝑠 𝑜𝑓 𝑓𝑖𝑛𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡

∑ 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑛𝑜𝑛 − 𝑏𝑒𝑛𝑖𝑔𝑛 𝑟𝑒𝑎𝑔𝑒𝑛𝑡𝑠 × 100 % X 100%

(Hudlicky et al., 1999)

Water intensity

WI

∑ 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑤𝑎𝑡𝑒𝑟

𝑚𝑎𝑠𝑠 𝑜𝑓 𝑓𝑖𝑛𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡 kg/kg X X X 0

(Curzons et al., 2001)

40

Most of these reaction-related and process-related metrics have the big advantage of being simple to use

and can give instant feedback of a change made in the process or in the synthetic reaction, and therefore can

easily be used for modelling and comparison of different scenarios. Thus there is a great interest in using

environmental metrics in the process development stage. However, it obviously also presents some

drawbacks and limitations. One evident limitation is the fact that the same weight is given to all different

types of waste generated. For instance, when calculating E-factor the same weight is given to 1 kilogram of

sodium chloride (NaCl) and to 1 kilogram of hexane, when clearly the health and safety hazards of both

chemicals is completely different resulting in different impacts on the environment (Tufvesson et al., 2012).

This is the reason why the calculation of these green chemistry metrics should be combined with an analysis

of the health and safety hazards of the chemicals used in the process.

In this thesis atom economy and carbon efficiency in combination with reaction mass efficiency were selected

for the comparison of the environmental profile of the different synthetic routes for (R)-perillic acid

production (section 3.3).

For the environmental assessment of an entire process a combination of several process-related metrics

were used in (R)-perillic acid and brazzein case-studies, but PMI was selected as the preferred metric to guide

process development.

2.2.2 Environmental, Health and Safety hazards

Assessing health and safety of the chemical inventories is an important part of an environmental assessment.

Besides human toxicity it is also important to take into account the hazards to the environment.

EHS aspects of the chemicals should be assessed based on the globally harmonized system of classification

and labelling of chemicals (GHS) and certain chemical hazard categories should be avoided (e.g. carcinogens,

mutagens, endocrine disrupting substances, flammable, explosive, etc.).

In addition to the health and safety of the chemicals, acute and chronic toxicity data should also be gathered

in response to legislation such as REACH. “Registration, Evaluation, Authorization and Restriction of

Chemicals” (REACH) is a regulation of the European Union, adopted to improve the protection of human

health and the environment from the risks that can be posed by chemicals, while enhancing the

competitiveness of the EU chemicals industry. It affects the import and usage of a wide range of chemicals

in Europe, and any products found to not comply with the conditions established in REACH are removed from

41

the market through the “Rapid Alert System for Dangerous Non-food Products” (RAPEX) information scheme

(Byrne et al., 2016). As a result of REACH, in 2015 several products containing toluene, chloroform or benzene

have been banned from EU, and several solvents are currently under scrutiny, such as N,N-

dimethylformamide (DMF), N,N-dimethylacetamide (DMAc), N-methylpyrrolidinone (NMP) as well as certain

hydroxyethers and chlorinated solvents (Byrne et al., 2016).

There are highly respected and trustworthy databases which compile information on the hazards of chemical

substances and help with the often complicated task of substituting a certain chemical. Examples of these

databases are the SIN LIST and the SUBSPORT, the Substitution Support Portal. The SIN (Substitute it Now!)

List is a globally used database of chemicals likely to be banned or restricted in a near future. The chemicals

on the SIN List have been identified by ChemSec (International Chemical Secretariat) as Substances of Very

High Concern based on the criteria established by the EU chemicals regulation REACH (http://sinlist.org,

accessed on 22.09.2016). SUBSPORT, the Substitution Support Portal, is a database/search engine on

hazardous substances that are legally or voluntarily restricted or subject to public debate (www.subsport.eu,

accessed on 22.09.2016).

These portals aim to support companies in fulfilling substitution requirements within EU legislation, as well

as being a useful resource for scientific institutions interested in substituting hazardous chemicals. It is

imperative that researchers look into the health, safety and environmental hazards of the chemicals used in

a certain process in an early stage of development, since substituting chemicals in later stages of research

and development is much more costly and complicated.

2.2.3 Life Cycle Assessment

In comparison with green chemistry metrics, Life Cycle Assessment (LCA) is at the other end of the spectrum

in terms of complexity and comprehensiveness (Lima-Ramos et al., 2014b). LCA is defined as “the compilation

and evaluation of the inputs, outputs and potential environmental impacts of a product system throughout

its life cycle” (European Comission for Standardization, 2006a). Life cycle thinking is an approach covering all

life cycle stages of a process, product or service system, from extracting and processing of raw materials,

manufacturing, transportation and distribution, use/reuse, recycling and disposal. This is often called a

“cradle to grave” approach, represented in Figure 2.5.

42

Figure 2.5 Life cycle thinking - a cradle to grave approach.

Unlike green chemistry metrics, LCA is a methodology that is not specific for chemical and/or biochemical

processes. It was developed to be suitable for the environmental assessment of all types of products,

processes or services (Lima-Ramos et al., 2014b).

LCA is a methodology that has been standardized by the International Standard Organization (ISO) –

ISO14040:2006 and ISO14044:2006 (European Comission for Standardization, 2006a) (European Comission

for Standardization, 2006b). The level of detail and the boundaries of an LCA study depend on the subject

and goal of the study, but the ISO standards need to be followed. The LCA methodology comprises of four

phases. The first phase is the goal and scope definition, where the system boundaries and goal of the study

are defined. This step is followed by the Life Cycle Inventory (LCI). This is the most time-consuming step and

consists of performing mass and energy balances to all the different steps in the product life cycle, in order

to quantify the material and energy inputs into the system as well as wastes and emissions from the system.

Then comes the Life Cycle Impact Assessment (LCIA) where all the inventory data is processed and aggregated

into a limited set of recognized environmental effects. The inputs and outputs quantified during the LCI step

are related to different impact categories, which include Global Warming Potential (GWP), Stratospheric

Ozone Depletion Potential (SOP), Photochemical Ozone Creation Potential (POCP), Acidification Potential

(AP), Nutrient Enrichment Potential (NEP), Eco-toxicity Potential (ETP), Human Toxicity Potential (HTP) and

Hazardous Waste Potential (HWP) (Lima-Ramos et al., 2014b). The last step consists of the interpretation of

Raw materials extraction

Transportation

Manufacturing

PackagingTransportation

Use

Disposal

43

the results obtained within the Life Cycle Inventory and/or Life Cycle Impact Assessment in the light of the

goal and scope defined. This last step typically includes a sensitivity and/or uncertainty analysis of the results

obtained in order to draw conclusions and make recommendations (Lapkin and Constable, 2009).

There are multiple commercially available software tools for the implementation of the LCA methodology,

e.g. SimaPro or GaBi. The biggest asset of these software tools is the inclusion of comprehensive inventory

databases, which facilitate the mass and energy balances of up- and downstream stages of the process under

investigation. One of the most comprehensive databases is the ecoinvent database (latest version is

ecoinvent 3.3) (www.ecoinvent.org accessed on 20.09.2016).

LCA provides a framework for reporting the environmental impacts reflecting the whole life cycle of a given

product and is therefore a valuable tool to support the implementation and marketing of a given product.

However, on the research and development stage of a process or product, a full LCA is too time-consuming,

laborious and complicated. The methodology is too comprehensive and there are difficulties in finding all the

necessary inventory data, which makes its applicability reserved to late stages of development when all this

information and data is certain and available. For these reasons, LCA does not enjoy widespread practice for

environmental assessment of bioprocesses, with the exception of biofuels (Tufvesson et al., 2012).

Given the complexity of the LCA methodology, several chemical companies have developed simplified in-

house LCA methods, based on the guidelines in the ISO standards for evaluating the sustainability of its

products. Examples of these methods are the FLASC-tool developed by GlaxoSmithKline to systematically

evaluate synthetic organic reactions and processes (Curzons et al., 2007), or the Eco-Efficiency Analysis

method developed by BASF (Shonnard et al., 2003).

2.3 Concluding remarks

There is a large number of available tools to assess the economic and environmental profile of a given

bioprocess and the selection of the most appropriate ones is not a straightforward task.

Regarding the assessment of the economic feasibility of a given process, process metrics such as product

concentration, space-time yield and reaction yield can provide meaningful information while being simple

and easy to calculate.

Similar metrics also exist for the environmental assessment of processes in early stages of development.

These metrics, commonly designated as Green Chemistry metrics, can be grouped into reaction-related or

44

process-related metrics and provide a quick assessment of the environmental profile of a chemical or

biochemical process.

Given the inherent differences in the processes, different tools should be used to evaluate biocatalytic and

fermentation processes, which are further explored throughout the thesis.

45

Part II

Case-studies

46

47

3. Case-study I: Bioconversion of (R)-limonene to (R)-perillic

acid by a whole-cell biocatalyst

3.1 Introduction

In this first chapter the work done for the first case-study is presented, including benchmarking of the

different routes for perillic acid production, economic and environmental assessment and development of

different improvement strategies for the process.

This case-study corresponds to the bioconversion of (R)-limonene to (R)-perillic acid by Pseudomonas putida

GS1. Perillic acid is a monoterpenoic acid that is naturally present in some plants of the genus Perilla, namely

in Perilla frutescens. It has a strong growth-inhibitory effect on bacteria and fungus, which makes it an

attractive candidate to be used for preservation purposes, e.g. in cosmetic industry (Mirata et al., 2009) (Rieks

et al., 2004). It has also been shown that perillic acid has anti-cancer properties, namely as an inhibitor of

metastic progression of melanoma cells, and could potentially be of interest to the pharmaceutical industry

(von Kitzing et al., 1994) (Yeruva et al., 2007).

The bioprocess for (R)-perillic acid production is presented in section 3.2. Section 3.3 corresponds to the

benchmarking of the bioprocess with other routes for perillic acid production. Section 3.4 corresponds to an

economic and environmental assessment of the bioprocess from both an economic and an environmental

perspective. Section 3.5 includes an experimental characterization of the process and section 3.6 a discussion

of the possible process technologies that can be used to improve this bioprocess.

3.2 Bioprocess description

(R)-Limonene is converted to (R)-perillic acid in a three-steps reaction represented in Figure 3.1. In the first

step (R)-limonene is converted to (R)-perillyl alcohol by a monooxygenase, which is then converted to (R)-

perillyl aldehyde by an alcohol dehydrogenase, and finally the aldehyde is converted to (R)-perillic acid by an

aldehyde dehydrogenase. This reaction is part of the degradation pathway for p-cymene that is naturally

expressed in Pseudomonas putida GS1 metabolism. This degradation pathway is an adaptation mechanism

that is induced in the presence of toxic solvents. Limonene is the monocyclic monoterpenoic analogue to the

48

aromatic p-cymene and it also activates the p-cymene degradation pathway. In this degradation pathway, p-

cymene is oxidized to p-cumate in three steps, which are catalyzed by a monooxygenase, an alcohol

dehydrogenase and an aldehyde dehydrogenase, which are also active towards limonene and its derivatives,

perillyl alcohol and perillyl aldehyde. But in contrast to p-cumate, perillic acid is not further catabolized and

it is secreted into the medium as a result of the narrow substrate specificity of the cumate dioxygenase that

further degrades p-cumate (Speelmans et al., 1998).

Figure 3.1 Bioconversion of (R)-limonene to (R)-perillic acid by P. putida GS1: (A) monooxygenase; (B) alcohol dehydrogenase; (C) aldehyde dehydrogenase.

As most organic solvents (R)-limonene is toxic to most bacterial strains, but Pseudomonas putida GS1 is able

to grow with concentrations of (R)-limonene as high as 500 mM since solvent tolerance mechanisms are

activated. On the other hand, (R)-perillic acid inhibits cell growth and the biocatalytic reaction and feedback

inhibition was claimed to determine the maximum product concentration achievable in batch cultures

(Speelmans et al., 1998). Previous work by Mirata and co-workers determined the critical inhibitory

concentration of (R)-perillic acid to be 165±7 mM. Product inhibition was confirmed experimentally and total

inhibition of growth of P. putida GS1 was observed with a concentration of 167 mM of (R)-perillic acid

(cultivations in TB media at pH 10). (Mirata et al., 2009).

A process for biotechnological production of (R)-perillic acid with a wild-type strain of Pseudomonas putida

GS1 was developed and established at BRAIN AG. Since the reaction system faces product inhibition, there is

the need of having an in situ product removal (ISPR) step to alleviate product inhibition.

49

3.3 Benchmarking of different routes for perillic acid production

In the first place, an evaluation of the environmental profile of different routes for perillic acid production

was made, with the purpose of identifying the most promising route for industrial production of perillic acid.

In the case of (R)-perillic acid all the three possible production routes represented in Figure 1.1 – extraction

from natural sources, chemical synthesis and biosynthesis – are theoretically possible. The routes for (R)-

perillic acid production that were identified in literature are: i) extraction from the plant Perilla frutescens; ii)

chemical synthesis starting from β-pinene, and iii) biocatalytic conversion of (R)-limonene to (R)-perillic acid

by Pseudomonas putida GS1.

Perillic acid has been identified as a monoterpene glucoside in the plant Perilla frutescens (Fujita and

Nakayama, 1993). However, it is only found in nature in very low concentrations as most terpenoids. There

are several oxyfunctionalised monoterpenes and sesquiterpenes which are industrially interesting to be used

as flavor and fragrances. But they are present in the essential oils of plants, which constitute around 0.1 to

5% of the plant. In the specific case of Perilla frutescens leaves, they are constituted by 1.04 %(w/w) of

essential oil (Yu et al., 1997). The extraction of minor compounds, such as perillic acid, from the essential oils

is only in very rare cases economically feasible (Berger, 2007). For this reason and since it was not found in

the literature the exact amount of perillic acid in the leaves of Perilla frutescens, this route for perillic acid

production was not included and it was not further analyzed.

On the other hand, terpene hydrocarbons such as (R)-limonene and pinenes are abundantly available in

nature as the main component of citrus and turpentine oils, respectively. Orange oil contains more than 90%

of (R)-limonene and turpentine oil up to 75-90% pinenes, which makes these compounds as an interesting

starting material for oxyfunctionalizations that lead to industrially relevant terpenoids, such as perillic acid

(Berger, 2007).

The biocatalytic conversion of (R)-limonene to (R)-perillic acid by Pseudomonas putida GS1 (Mars et al., 2001)

(Speelmans et al., 1998) (Mirata et al., 2009) described in section 3.2 and the chemical synthesis of perillic

acid starting from β-pinene (Wang et al., 1993) were therefore compared from an environmental point of

view.

The biocatalytic conversion of (R)-limonene to (R)-perillic acid by Pseudomonas putida GS1 is shown in Figure

3.1.

The chemical synthesis of perillic acid starting from β-pinene is shown in Figure 3.2.

50

Figure 3.2 Chemical synthesis of perillic acid starting from β-pinene (Wang et al., 1993).

Based on the mass balances of both routes, the key reaction-related green chemistry metrics were calculated

for both synthetic routes and are shown in Table 3.1. For easier interpretation, Figure 3.3 shows a graphic

representation of these green chemistry metrics for both synthetic routes.

Table 3.1 Comparison of reaction-related green chemistry metrics for the biocatalytic and chemical synthetic routes for perillic acid production.

Biocatalytic route Chemical route

Atom Economy (%) 99 25

Carbon Efficiency (%) 100 59

Reaction Mass Efficiency (%) 20 3

51

Figure 3.3 Reaction-related green chemistry metrics for the biocatalytic and chemical synthetic routes for perillic acid production.

From Table 3.1 and Figure 3.3 it is clear that the biocatalytic route has the potential to be a “greener” process,

since all the three metrics are considerably superior to the chemical route. Carbon efficiency is 100%, which

corresponds to the ideal value, and atom economy is also extraordinarily close to the ideal. The starting

material for the biocatalytic route, (R)-limonene is chemically very similar to the final product, (R)-perillic acid

and only oxygen is consumed as a reactant in this oxyfunctionalization. Therefore, this is a very good example

of a synthetic route that obeys to the principles of atom economy – the synthetic route maximizes the

incorporation of the starting materials into the final product.

However, the situation is a little different when it comes to RME. This metric also has a higher value for the

biocatalytic route, but in both routes its value is very low which indicates that both synthetic routes require

further development and optimization.

Another important issue to take into account when selecting a synthetic route is the qualitative analysis of

EHS aspects of the chemicals and solvents used in the reaction. McElroy and co-workers grouped several

solvents that are typically used in the pharmaceutical industry and attributed green, yellow or red flags to

the use of hazardous and highly hazardous solvents, problematic and recommended solvents, respectively.

The same procedure was adopted for the safety and health of the reactants used, based on the H-statements

defined by the globally harmonized system of classification and labeling of chemicals (GHS) (McElroy et al.,

2015). This flags system is very simple, illustrative and easy to interpret and has therefore been followed to

compare the chemical properties of the chemicals used in both routes.

0

10

20

30

40

50

60

70

80

90

100

AE (%) CE (%) RME(%)

Biocatalytic route Chemical route

52

Table 3. 2 Analysis of the environmental, health and safety hazards of the reactants and solvents used in the biocatalytic and chemical synthetic routes.

Green flag Yellow flag Red flag

Biocatalytic route

Solvents No solvents used. X

Reactants (R)-Limonene H410: Very toxic to aquatic life with long lasting effects

X

Chemical route

Solvents Nitromethane – highly hazardous X Hexane - hazardous X Methanol X Water X

Reactants Hydrogen peroxide (H2O2) H330: Fatal if inhaled

X

Manganese dioxide (MnO2) H373: May cause damage to organs through prolonged or repeated exposure

X

Analyzing Table 3. 2 one can see that the chemical synthesis shows more EHS concerns (translated by the

number of red and yellow flags) than the biocatalytic route. The chemical synthesis consists of four different

steps, each one with different reactants, solvents and catalysts. So in total, the number of chemicals used in

the chemical route is considerably higher than the biocatalytic route. Two of the solvents used in the chemical

route are hazardous. Nitromethane is highly hazardous, since it is a flammable liquid and vapour (H226) and

highly explosive, and hexane is also a hazardous solvent which shows several health, safety and

environmental issues – it is a highly flammable liquid and vapour (H225), it is suspected of damaging fertility

(H361), may cause damage to organs through prolonged or repeated exposure (H373) and is toxic to aquatic

life with long lasting effects (H411). Hexane has also been considered a chemical of very high concern by

ChemSec and it is part of the SIN List because it is classified as a possible reprotoxic chemical (R3) and has

been reported to have endocrine and neurological effects (sinlist.org accessed on 18.09.2016). The reactants

and catalysts also show EHS hazards, for instance manganese dioxide (MnO2) may cause damage to organs

through prolonged or repeated exposure (H373). Another environmental issue of the chemical route is the

fact that each synthetic step takes place at a different temperature in a range from 0 °C to 80 °C, which is

undesirable from an energetic point of view.

On the other hand, the biocatalytic route also shows some EHS concerns. (R)-limonene is very toxic to aquatic

life with long lasting effects (H410) and it is therefore not the “greenest” starting material. Moreover, when

53

making the chemical inventory of the biocatalytic route, CoSO4 has been identified as a chemical of very high

concern, present in the SIN List since it has been classified as carcinogenic, mutagenic, reprotoxic (CMR)

according to Annex 1 of Directive 67/548/EE (sinlist.org accessed on 20.08.2016). However, cobalt sulphate

is only used in the trace metals solution and it is present in the reaction media in a concentration in the range

of nanograms per liter. Since, the concentration of cobalt sulphate used is so low, this should not be a major

health and safety concern. When assessing the EHS hazards of the different chemicals used in a chemical

reaction it is also important to analyze the EHS hazards of the final product. However that was not possible

since data on the safety and toxicology of (R)-perillic acid could not be found in scientific literature.

The use of a biocatalyst is preferable from an environmental point of view, as opposed to the use of excess

of reagents in the chemical route. And the fact that the reaction takes place at a constant temperature close

to room temperature (30 °C) is also a positive point from an environmental and an economic perspective.

In conclusion both the reaction-related green chemistry metrics and the EHS hazards of the chemical

inventory point towards the selection of the biocatalytic route as the most promising to be investigated for

industrial application. The economic and environmental profile of the biocatalytic reaction were further

investigated in the following sections of the thesis.

3.4 Bioprocess evaluation

3.4.1 Economic assessment

3.4.1.1 Process metrics

In order to make an initial assessment of the economy of the bioprocess for (R)-perillic acid production, the

process metrics that are relevant in this case, i.e. product concentration, space-time yield and reaction yield,

were calculated based on the mass balances of the process. The process metrics for are shown in Table 3.3.

54

Table 3.3 Process metrics for the bioprocess for (R)-perillic acid production. The reaction yield is calculated on a molar basis.

Process metrics (R)-perillic acid bioprocess

Product concentration (g/L) 4.81

Space-time yield (g/(L.h)) 0.17

Reaction yield (%) 20.6

Figure 3.4 Process metrics for the (R)-perillic acid bioprocess presented along with typical values for an economically feasible process.

Characterization and identification of the limitations of a bioprocess is the first step towards development of

an economically feasible process. Translating the performance of the bioprocess into process metrics and

benchmarking against quantitative targets is a very valuable tool. In Figure 3.4 typical values for a biocatalytic

pharmaceutical process implemented at industrial scale are represented together with the process metrics

for the bioprocess for (R)-perillic acid production. Pharmaceutical chemicals can be used as an appropriate

benchmark given the low annual production that is expected for (R)-perillic acid. Since it is a novel chemical

product that is not commercially available in the cosmetic, food, pharmaceutical or any other market, it is

assumed a very low initial annual production of (R)-perillic acid in the range of 1 ton/year.

As shown in Figure 3.4 all the three process metrics for the bioprocess for (R)-perillic acid production are far

from reasonable values for an economically feasible process. The process undoubtedly needs further reaction

and process development for its intensification and it is still far from economic feasibility.

4,81

0,17

20,6

20

1

90

Product concentration (g/L)

Space-time yield (g/(L.h))

Reaction yield (%)

Industrial processes (R)-perillic acid bioprocess

55

Furthermore, all of the three process metrics are approximately equally far from the targets, i.e. none of

them is significantly better than the others. And all the process metrics are equally important for the

economy of a biocatalytic process. So the question that is raised from this simple economic assessment is –

which metric should be improved in first place? Which metric has a greater influence on the overall economy

of the process and should therefore be tackled first?

In order to further explore this question a process model and a full cost estimation were developed which

are shown in section 3.4.1.2.

3.4.1.2 Process model

Economic feasibility was assessed by means of a method designed for utilization at early development stages

of chemical processes (Heinzle E.; Weirich and K., 1998) (Heinzle et al., 1998). A process model for the

chromatography-based process was developed in the commercial process simulation software SuperPro

Designer®, version 9.5 (Intelligen Inc., Boston, USA).

The model is based on experimental data collected at 10L scale and scalability to 5 m3 was assumed.

The bioconversion reaction was described as in equation Equation 3.1 and a reaction yield of 20.6% was

assumed. It was assumed that the remaining (R)-limonene that is fed to the fermenter is lost in the exhaust

gas, a phenomenon that is observed at laboratory scale.

Equation 3.1

𝐶10𝐻16 + 1.5 𝑂2 → 𝐶10𝐻14𝑂2 + 𝐻2𝑂

The annual production was assumed to be 1 ton (R)-perillic acid per year and the annual production time is

only 3 months/year. It is assumed that the final product would be stored in storage facilities for the remaining

time of the year.

The scheduling of the process was done in order to maximize equipment usage. Total batch time is

approximately 139 hours (6 days) and 17 batches run in one production year.

56

SuperPro Designer® software comprises comprehensive databases for the calculation of numerous economic

parameters. These values were used for the economic assessment unless otherwise stated. Raw material

costs were complemented with industrial market prices from the ICIS webpage (www.icis.com accessed on

15.07.2016) or, if necessary, by division of the lab-scale prices from Sigma-Aldrich (www.sigmaaldrich.com

accessed on 20.10.2016) by a factor of 10.

Basic labor costs are estimated from the labor requirements of each operation step in the factory and

assuming a standard wage of 30 $/h (approximately 28 €/h) for an operator (excluding benefits) (Kuhn et al.,

2010). It was also assumed that the industrial facility would be a multi-purpose plant with shared utilities and

services, which would be rented for (R)-perillic acid production. The facility-dependent costs, which

represent a rental fee for the use of the equipment, are calculated based on equipment usage. Wastewater

treatment costs were not included in the cost assessment.

Based on this process model, which includes mass as well as energy balances, an economic assessment of

the process including a full cost estimation was made.

The economic assessment that was performed focused solely on the production cost (€/kg product). Capital

cost (CAPEX) and depreciation of the equipment were not considered in the economic assessment.

The results of the economic assessment revealed that the production costs are higher than what would be

desired, which confirmed the previous conclusions from the assessment of the process metrics that the

process needs to be further intensified.

Figure 3.5 shows the contribution of different elements to the total annual operation cost. These different

elements are raw materials, labor, facility-dependent costs, utilities and consumables, which include resins,

filters, etc.

57

Figure 3.5 Annual operation cost breakdown for the chromatography-based production process.

Facility-dependent costs are the biggest contributor to the total operations costs, which is not uncommon

for bioprocesses. Bioprocesses typically produce high-value products in small quantities in relatively

expensive facilities, with equipment that is more sophisticated, e.g. fermenters and chromatographic

columns, than equipment that is used in conventional chemical processes (Harrison et al., 2015). That is

precisely the case here where (R)-perillic acid is produced in a very small scale (1 ton/year), therefore it is

expectable that facility-dependent costs contribution to total operation cost is substantial.

Biocatalytic processes are typically capital intensive because large and complex fermenters are required. The

main reasons for this are usually dilute reaction systems and complex requirements for process control to

guarantee optimal growth of microorganisms (Straathof and Adlercreutz, 2000). Sterilization and

biocontainment are also extra features of fermenters (when compared with chemical reactors) that result in

an additional cost. The comparably dilute product concentrations in the fermenter are the reason why the

equipment has to be designed larger as compared to chemical plants to achieve similar product amounts.

This results in a higher cost of the equipment, which can be alleviated if the equipment is rented, as it

happens in this case. The considerably high facility-dependent costs stress the importance of process

intensification by different process technologies and/or biocatalyst modification. (Woodley, 2008) (Chotani

et al., 2000).

Raw materials

5%Labor17%

Facility-Dependent

61%

Consumables 17%

Utilities 0%

58

(R)-perillic acid production at a 1 ton/year scale was clearly dominated by fixed costs (facility-dependent and

labor costs), which account for 78% of the total operation costs. That is a similar scenario to other published

economic assessments of biocatalytic processes, e.g. (S)-styrene biocatalytic production where fixed costs

account with 63% of total operations costs (Kuhn et al., 2010) (Kuhn et al., 2012).

The high fixed costs and the low variable production costs (raw materials and utilities) of the biocatalytic

process would clearly favor larger annual production volumes, as the contribution of the fixed costs with

respect to total costs decreases with increasing scale (economy of scale). This was confirmed by running a

sensitivity analysis on the volume of annual production, whose results are shown in Figure 3.6.

Figure 3.6 Relative unit production cost (€/kg) as a function of annual production (ton/year).

But more interesting than merely calculating the production cost and the weight of its different contributors

is to analyze what influences it. Therefore a sensitivity analysis on the different process metrics was made

and the variation of production cost was analyzed.

The sensitivity analysis was performed by analyzing the influence of reaction yield and reaction time

(expressed by the space-time-yield) on unit production costs (€/kg).

For the sensitivity analysis on reaction yield, the annual production was kept constant at 1 ton/year and the

results are shown in Figure 3.7.

0

20

40

60

80

100

0 1 2 3 4 5

Rel

ativ

e u

nit

pro

dct

ion

co

st (

%)

Annual Production (ton/year)

59

Figure 3.7 Relative production cost as a function of increase in reaction yield.

Increasing the reaction yield up to 4 folds, only decreased the production cost by 0.5%, which is an

insignificant difference from an economic perspective. This minor decrease can be explained by the substrate

low selling price. By increasing the reaction yield, the raw materials cost contribution to total operation costs

is reduced. But since the substrate is a low-value chemical, the cost contribution of the raw materials is

anyway low (around 5% of total operation costs as shown in Figure 3.5). So increasing reaction yield does not

have a significant impact on total operation costs.

On the other hand, the sensitivity analysis to the space-time yield revealed a different picture. Figure 3.8

shows the results of this sensitivity analysis, where space-time yield was varied by only changing the reaction

time (reaction yield was kept constant). The results show that the production cost is very sensitive to space-

time yield. For instance, when reducing transformation time by 80%, total operation costs can be reduced up

to 40%, which is a very significant reduction.

In this sensitivity analysis, annual production was also kept constant as 1 ton/year. The reduction in

production cost can be explained by the fact that if the transformation time is reduced, the total batch time

is also reduced. Hence, to produce 1 ton of (R)-perillic acid a lower annual production time is necessary, which

reduces the cost of equipment usage and labor. Since the facility-dependent costs are the biggest contributor

to total production costs, reducing the reaction time can have a very significant effect on the economy of the

process.

60

Figure 3.8 Relative production cost and increase in space-time yield as a function of reduction in transformation time.

In conclusion, the sensitivity analysis on different process metrics allowed the identification of the space-

time yield, or more specifically the reaction time, as one of the parameters that mostly affect the economic

feasibility of this process. Therefore, the focus of process development should be on increasing space-time

yield and accelerating the reaction.

The process model and subsequent full cost estimation proved to be a very useful tool in ranking the

importance of the different process metrics and identifying the space-time yield as the process metric that

mostly affects the economy of this bioprocess.

3.4.2 Environmental assessment

The environmental profile of the biotechnological production of (R)-perillic acid was also assessed in terms

of process-related green chemistry metrics. The green chemistry metrics (PMI, E-factor and SI) for the

bioprocess for (R)-perillic acid production were calculated based on experimental data at 10L scale and are

shown in Table 3.4.

61

E-factor of the bioprocess for (R)-perillic acid production is relatively high, but within the typical range of

values for pharmaceutical chemicals (25 to >100). PMI is also very similar in both cases and relatively high,

which again demonstrates that the process needs further intensification.

Table 3.4 Process-related green chemistry metrics for the bioprocess for (R)-perillic acid production.

Process-related metrics (R)-perillic acid

bioprocess

SI (kgsolvent/kgproduct) 109.8

E-factor (kgwaste/kgproduct) 226.9

PMI (kgmaterials/kgproduct)

USP 83.1

DSP 219.8

Total 302.8

Figure 3. 9 shows the contribution of different materials to total PMI of the bioprocess for (R)-perillic acid

production.

Figure 3. 9 Composition by mass of the type of materials used in the bioprocesses for (R)-perillic acid production.

Water57%

Reactants5%

Solvents37%

Others1%

62

The biggest contributor to PMI is water, which is typical of bioprocesses. Solvents contribute with 37% to PMI

due to ethanol consumption in the downstream processing section of the bioprocess. Even though ethanol

is not a hazardous solvent, solvents use should always be minimized.

3.5 Experimental characterization of the bioprocess

In the previous sections an economic and environmental assessment of the bioconversion of (R)-limonene to

(R)-perillic acid was done. One of the conclusions from the economic assessment was that space-time yield

or volumetric productivity (g/L/h) greatly affects the economy of the process.

Here, as an addition to the in silico analysis of the bioprocess, some experiments to characterize it were done

in order to investigate the reasons for the low space-time yield and better understand the limitations of this

reaction system.

3.5.1 Materials and methods

3.5.1.1 Determination of the reaction rate

To determine the reaction rate of (R)-limonene conversion to (R)-perillic acid, a fermentation following the

same process conditions as the ones followed by BRAIN AG was run.

Chemicals, strain and media

(R)-limonene (99% purity) was purchased from Erich Ziegler (Aufsess, Germany), (S)-perillic acid (95% purity)

and all other chemicals have been purchased from Sigma-Aldrich (Seelze, Germany).

P. putida GS1 was purchased from DSMZ (German Collection of Microorganisms and Cell Cultures,

Braunschweig, Germany). Minimal media with glycerol as the main carbon source was used both for pre-

cultivation and cultivation in the fermenter.

63

Fermentation

The bioconversion of (R)-limonene to (R)-perillic acid was performed according to standard procedures

(Mirata et al., 2009).

Analytical methods

The concentration of (R)-perillic acid was measured by HPLC analysis. Samples from the fermentations were

centrifuged at 13,000 rpm for 10 minutes and the supernatant was diluted in an equal volume of acetonitrile.

The sample was again centrifuged at 13,000 rpm for 10 minutes for protein removal and further diluted with

acetonitrile until a concentration that fits the calibration curve was reached (calibration curve shown in

Appendices). Commercially available (S)-perillic acid (Sigma-Aldrich, Germany) was used as an external

standard, since commercially available (R)-perillic acid could not be acquired.

The HPLC analysis was performed on a Dionex UltiMate 3000 HPLC equipped with a Photodiode Array

Detector (Dionex, Thermo Scientific). The column used in the HPLC was a Luna 3μm C18(2) 100 Å, 150x4.6

mm. The mobile phase was 70:30 (v/v) acetonitrile/water containing 0.1% formic acid, at 0.8 mL/min and 40

°C. The UV detector wavelength was selected at 217 nm.

Measurement of cell concentration

To measure growth kinetics, cell concentrations were analyzed photometrically at 600 nm (OD600nm). Before

measuring cell concentrations in limonene conversions, 1 mL samples were centrifuged at 13,000 rpm for 10

minutes and cell pellets were resuspended in a sterile saline solution (0.9 %(w/v) NaCl) after removing the

supernatant and limonene phase.

For determination of cell dry weight, 1 mL of cell culture was transferred into a pre-weighed and pre-dried

Eppendorf tube. The sample was then centrifuged at 13,000 rpm for 10 min in an Eppendorf centrifuge. The

supernatant was carefully extracted and the cell pellet was washed with a sterile saline solution 0.9 %(w/v)

and centrifuged again (13,000 rpm; 10 min). The cell pellet was then dried at 80°C for more than 24 hours

and placed in a desiccator to cool to room temperature until a constant weight at four decimal places was

measured.

64

3.5.1.2 Measurement of substrate loss under fermentation conditions

Limonene loss under fermentation conditions was measured in two different sets of experiments. In the first

set, aeration rate was fixed and limonene loss in the exhaust gas was measured over time. In the second set

of experiments, the aeration rate was varied using a flow control valve and limonene loss was measured for

a fixed period of time.

These measurements were made in an experimental setup simulating the fermentation conditions but

without cells. The temperature was kept constant at 30 °C, agitation was constant at 600 rpm and 2 %(v/v)

of limonene was added to the fermentation media at the beginning of the experiment. The experimental

setup, represented in Figure 3.10, consisted of a reactor vessel with an incorporated sparger at the bottom

(a ring sparger, the same type of spargers used in the fermenter). The off-gas was collected in a Viton tube,

which is impermeable to limonene and condensed using dry-ice as the cooling agent. For the experiments

with variable aeration rate, the air flow was controlled and measured with a flow control valve. Aeration

rates between 200 mL/min and 1000 mL/min were tested.

To measure the remaining limonene in the fermenter after a certain period of time, limonene was extracted

from the fermentation media with methyl tert-butyl ether (MTBE) and measured in a UV-VIS

spectrophotometer (Shimadzu, Japan). It was not possible to take reproducible samples from the reactor

vessel since limonene is poorly-soluble in water (solubility in water is 0.1 mM) and droplets of limonene are

dispersed in the aqueous media. Therefore, the entire volume of media was used in a liquid-liquid extraction

with MTBE to collect the remaining limonene in the fermenter after a certain time of operation. UV-VIS

measurements were possible because limonene is slightly yellow. Different wavelengths were tested and it

was seen that the UV-VIS absorption spectrum of MTBE and glycerol do not overlap the limonene absorption

peak. A calibration curve at 254 nm was made for concentrations of limonene in the range of 0.1-2 %(v/v)

limonene.

65

Figure 3.10 Experimental setup for experiments measuring limonene stripping from the fermenter.

3.5.2 Results

3.5.2.1 Determination of the reaction rate

To determine the rate of the conversion of (R)-limonene to (R)-perillic acid the reaction was followed for 48

hours. The progress curve for the reaction is shown in Figure 3.11. and Figure 3.12 shows the progression of

biomass growth and (R)-perillic acid formation for the entire time frame of the experiment.

Reactor vessel

Condensor with Dry Ice

Gas outlet

Air inlet

Flow control valve Flow measurement

66

Figure 3.11 Progress curve of the conversion of (R)-limonene to (R)-perillic acid in 1 L fermenter. A pulse of (R)-limonene was added to the fermenter at time 0 h and a second pulse was added after 24 hours of

reaction (represented by the grey line).

Figure 3.12 Biomass growth ( ) and product formation (●) kinetics in 1 L fermenter. The grey lines represent the time of (R)-limonene addition (first addition 17h after inoculation and the second addition is

24h after the first addition).

In Figure 3.11 one can observe that after (R)-limonene addition to the fermenter there is a lag phase of

approximately 3 hours where the production of (R)-perillic acid is negligible. This suggests that some time is

0

5

10

15

20

25

30

0 10 20 30 40 50

[(R

)-P

erill

ic a

cid

] (m

M)

Reaction time (h)

0

5

10

15

20

25

30

0

5

10

15

20

25

30

0 10 20 30 40 50 60 70

OD

60

0n

m

[(R

)-p

erill

ic a

cid

] (m

M)

Time (h)

67

required to activate the induction of cellular adaptation mechanisms in response to the toxic effect of (R)-

limonene. After this adaptation phase, the conversion of (R)-limonene to (R)-perillic acid takes place at a

given rate for the following 5 hours until a concentration of approximately 20 mM is reached. The reaction

reaches then a plateau and the concentration of (R)-perillic acid remains constant until the addition of a

second pulse of (R)-limonene. Then the concentration of (R)-perillic acid increases again but at a lower rate

than before and again reaches a plateau at a concentration of 25 mM.

Taking into account only the first 8 hours of reaction (after which the concentration of (R)-perillic acid remains

approximately constant) the conversion takes place at an average rate of 2.3 mM/h, which corresponds to

0.38 g/L/h. This value is slightly higher than the ones reported by Mirata and co-workers that report a

conversion rate of 1.8 mM/h (0.30 g/L/h) in the first 4 days of operation of the bioprocess developed by them

(Mirata et al., 2009).

The plateau of (R)-perillic acid concentration could be explained by either i) issues on substrate availability,

ii) product inhibition, or a combination of both problems.

Regarding substrate availability, it was easily detected that a significant part of (R)-limonene that is fed to

the fermenter is lost in the exhaust gas, since (R)-limonene has a characteristic smell to orange. Limonene

vapor pressure at 30 °C is approximately 3.4 mmHg (valor estimated from (Perry et al., 1997)), which is lower

than water vapor pressure at 30 °C (31.8 mmHg from (Perry et al., 1997)). Therefore, its volatility is lower

than water volatility under fermentation conditions. However, since limonene is poorly-soluble in water and

there is an inlet air stream, stripping of (R)-limonene occurs. (R)-Limonene is stripped by the gas stream and

lost in the exhaust gas. Hence, when the conversion of (R)-limonene to (R)-perillic acid reaches a stationary

phase, it could be because there is no more limonene available in the fermenter and most of it was lost in

the exhaust gas.

Regarding product inhibition, Mirata and co-workers have investigated product inhibition in this reaction

system and they observed that maximum specific activities (μmol perillic acid/min/g cdw) decreased

exponentially with increasing product concentrations. Already at (R)-perillic acid concentrations as low as 2

mM, the maximum specific activity of the biocatalyst decreases (Mirata et al., 2009). Therefore, at a

concentration of 20 mM of (R)-perillic acid the specific activity of the whole-cell biocatalyst is significantly

impaired, which can contribute to the reduction in conversion rate and the plateau on (R)-perillic acid

formation. After the second addition of (R)-limonene the conversion of (R)-limonene to (R)-perillic acid takes

place, but at a much lower rate which might indicate that the whole-cell biocatalyst’s activity has been

reduced due to a prolonged contact with the product.

68

But since the biocatalyst is a whole-cell there are several different factors influencing the conversion rate

besides the activity of the enzymes involved in the biosynthetic reaction. During the fermentation the cells

are under constant stress and its metabolism is overloaded. A high amount of energy is spent in activating

the solvent tolerance mechanisms to cope with (R)-limonene presence in the cultivation media, in the

conversion of (R)-limonene to (R)-perillic acid and finally in pumping (R)-perillic acid out of the cells.

Therefore, it is possible that with time the cells are no longer viable and able to properly function under these

conditions. The cell’s membrane might be compromised and the cells unable to pump (R)-perillic acid out of

their membrane, which would be translated in a reduction of the experimentally measured reaction rate.

The cells might also be unable to properly regenerate the cofactor necessary for the bioconversion, or it is

also possible that being unable to cope with all these stress factors, a large number of cells die which means

that in the end of the fermentation only a small number of cells is able to perform the bioconversion. Finally,

given the fact that the strain that is used is a wild-type strain it is also possible that a large number of cells

mutate and lose the p-cymene pathway, losing their ability to convert (R)-limonene to (R)-perillic acid.

The experimental determination of the reaction rate proved to be very useful for the development of the

process. In the bioprocess for (R)-perillic acid production, the ISPR operation takes place only 24 hours after

the addition of limonene. But according to the progression curve obtained in this experimental work, the

ISPR should take place much earlier and be more frequent. If a plateau is reached after 8 hours of reaction,

the remaining 16 hours until ISPR are merely wasted since there is not a significant improvement on product

formation. By keeping the reaction always at a steady rate and avoiding that the conversion reaches a

stationary phase, the space-time yield will be higher. This results in a more efficient use of the equipment,

namely the fermenter, and can further reduce facility-dependent costs that have been shown to be one of

the biggest cost contributors in this process.

3.5.2.2 Measurement of substrate loss under fermentation conditions

To further investigate the causes of a reduced reaction rate, (R)-limonene loss under fermentation conditions

has been investigated. (R)-Limonene loss in the exhaust gas stream affects negatively the reaction yield and

the conversion rate. Besides, (R)-limonene is very toxic to aquatic life with long lasting effects (H410), so it

should not t be released to the environment. Hence, experiments were run in order to quantitatively measure

how much (R)-limonene is lost during the conversion time.

69

In the first set of experiments the aeration time was fixed to 4 hours but the aeration rate was varied.

Whereas in the second set of experiments, the aeration rate was kept constant at 1 vvm for different periods

of time (4 hours and 12 hours). The results for both experiments are shown in Figure 3.13.

The experimental method showed some practical difficulties and limitations and some variability in the

measurements was observed (Figure 3.13). Given the variability of the experimental measurements it was

not possible to quantitatively measure the amount of (R)-limonene that is lost during the bioconversion.

But nonetheless, it was possible to observe and experimentally show that (R)-limonene stripping under

fermentation conditions takes place. As expected, the loss of limonene is more pronounced for higher

aeration rates and when the aeration runs for longer periods of time, i.e. as reaction time goes by more (R)-

limonene is lost in the exhaust gas. However, it was not possible to draw any conclusions on the extent that

aeration rate or aeration time influence limonene loss under fermentation conditions. Moreover, these

experiments were run in the absence of cells, but their presence could also influence limonene stripping.

Even though the accuracy of the experimental method does not allow an accurate quantification of limonene

loss under fermentation conditions, it clearly shows that stripping of the substrate does occur which confirms

that substrate availability is a major issue in this reaction system. Since (R)-limonene is stripped from the

fermentation media over time, it should be fed to the fermenter more frequently than only every 24 hours,

or possibly in a continuous mode. Given that (R)-limonene stripping is reduced at lower aeration rates, the

y = 2,2921x + 23,743R² = 0,8152

0

10

20

30

40

50

60

0 5 10

Lim

on

ene

lost

(%

)

Time (h)

y = 0,0187x + 5,3953R² = 0,5609

0

5

10

15

20

25

30

35

40

200 400 600 800 1000

Lim

on

ene

lost

(%

)

Aeration rate (mL/min)

Figure 3.13 Limonene loss under fermentation conditions. Left: Limonene loss at different aeration rates for a fixed period of aeration of 4 hours. Right: Limonene loss at different periods of time for a fixed aeration rate of 1 vvm.

70

possibility of running the reaction at lower aeration rates should also be considered. The current bioprocess

runs with growing cells, which means that oxygen is required both for cell growth and for the bioconversion

of (R)-limonene to (R)-perillic acid. Therefore, there is a limit of how much the aeration rate can be reduced

and alternative methods for oxygen supply, such as supplying enriched air or producing oxygen in situ by the

decomposition of hydrogen peroxide by catalase, should be considered. Another option to reduce (R)-

limonene stripping would be to run the reaction in a two-phase system. An organic solvent layer could act as

a (R)-limonene reservoir and in this way stripping could be reduced. This system has also the benefit of

potentially reducing substrate and product inhibition, but would require some modifications in the process

operation.

3.6 Strategies for improvement of the bioprocess

In previous sections was shown that space-time yield is the metric that affects the most the production cost.

After an experimental characterization of the process, it was observed a reduction of the conversion rate

approximately 8 hours after the start of the reaction. The production of (R)-perillic acid reaches a stationary

phase where there is no significant increase on its concentration. Investigation on substrate loss under

fermentation conditions showed that there are problems with substrate availability in the reaction. And

previous work from Mirata and co-workers showed that the reaction faces product inhibition from very low

concentrations of (R)-perillic acid (Mirata et al., 2009). All these issues affect negatively space-time yield.

Space-time yield is a very complex metric which is affected by several different parameters, summarized in

Table 3.5. Different strategies for improvement of these parameters, including tools for both biocatalyst

modification and process development, are shown in Table 3.6. These different parameters are reviewed and

discussed for whole-cell P450 processes by Lundemo and co-workers (Lundemo and Woodley, 2015).

71

Table 3.5 Typical parameters affecting space-time yield.

Biocatalyst-related parameters

Specific activity (U/g cdw)

Fermentation cell density (g cdw/L)

Enzyme activity

Cofactor regeneration

Electron transport from the cofactor to the active site

Substrate and product related parameters

Transport of substrate and product across the cell membrane

Product and/or substrate inhibition

Process related parameters Oxygen requirements

Transport limitations between phases in a two-phase system

Table 3.6 Different strategies to improve space-time yield.

Biocatalyst specific activity

(U/g cdw)

Fermentation cell density (g cdw/L) High cell-density fermentation

Fermentation optimization (media, feeding, etc.)

Enzyme activity Protein engineering

Cofactor regeneration

Change operating mode

Metabolic engineering

Protein engineering

Heterologous enzymatic regeneration of the cofactor

Electron transport from the cofactor to the active site

Engineering of redox partners

Substrate and product related

parameters

Transport of substrate and product across the cell membrane

Permeabilization of the cell membrane (physical or chemical methods)

Product and/or substrate inhibition In situ product removal (ISPR) In situ substrate supply (ISSS)

Substrate solubility Addition of co-solvents

Process-related parameters

Oxygen requirements Alternative methods for O2 supply

Transport limitations between phases in a two-phase system

Addition of surfactants or co-solvents

A whole-cell biocatalyst is a very complex system and therefore there are several different parameters

affecting space-time yield. Regarding the parameters related to the biocatalyst, it is the biocatalyst specific

activity (U/g cdw) that mostly affects the space-time yield. The biocatalyst specific activity is influenced by

the cell density in the fermentation, the enzyme activity, in terms of kcat (s-1, turnover per active site and

second), cofactor regeneration and electron transport from the cofactor to the active site. The bioconversion

of (R)-limonene to (R)-perillic acid is a multi-enzyme system, with three different enzymes involved in the

reaction (or four enzymes if the reductase shuttling NADH to the first reaction step is also included).

Therefore, for improvement of any of the biocatalyst-related parameters affecting space-time yield, such as

enzyme activity, cofactor regeneration or electron transport from the cofactor to the active site, it is

72

necessary to determine the rate limiting step and improve the enzyme(s) involved in that step. From a

process engineering point of view it is possible to improve the fermentation cell density by optimizing the

fermentation conditions, e.g. the fermentation media or the feeding of the carbon source. In order to

increase the fermentation cell density, a high-cell-density fermentation of Pseudomonas putida GS1 was

developed and is shown in the following sections.

Regarding the substrate and product related parameters, this process faces product inhibition and low

solubility of the substrate in water. One of the strategies to improve problems of product inhibition in the

reaction is to have ISPR. The ISPR operation of this process was also further investigated in the following

sections.

High-cell density fermentation and ISPR were further investigated and tested experimentally and are

presented as examples of process technologies that can be applied to intensify this bioprocess.

3.6.1 Materials and methods

3.6.1.1 High-cell-density fermentation

The seed culture and the fermentation ran in the same conditions as the ones applied in the base-case

fermentation. The only difference in this experiment was the feeding profile of the carbon source used in the

biomass growth phase. The feeding of 40 %(w/v) glycerol was done exponentially, according to Equation 3.2,

in order to keep a constant specific growth rate of 0.1 h-1. In Equation 3.2 μ0 corresponds to the specific

growth rate (h-1); YSX is the yield of biomass on carbon source; Sf is the concentration of carbon source in the

feed stream (g/L); S0 is the concentration of carbon source in the culture vessel at the end of the batch phase

(g/L); X0 is the biomass concentration in the culture vessel at the end of the batch phase (g/L) and V0 is the

initial culture volume, before the fed-batch has started.

Equation 3.2

𝐹(𝑡) =𝜇0

𝑌𝑆/𝑋 (𝑆𝑓 − 𝑆0) 𝑋0𝑉0 exp (𝜇0𝑡)

73

In these experiments, YSX was assumed to be 0.26 g cdw/g glycerol based on shake flask fermentations, the

initial culture volume (V0) was 700 mL, Sf is equal to 0.4 g glycerol/mL; it is assumed that the carbon source is

entirely consumed by the end of the batch phase (S0 is equal to zero) and X0 was measured after the batch

phase of 5 hours.

The exponential feed lasted 24 hours after which the feeding strategy was changed to a constant feeding.

(R)-Limonene was fed to the fermenter after the exponential feed in order to start the bioconversion of (R)-

limonene to (R)-perillic acid and the reaction was followed for 48 hours.

3.6.2 Results

3.6.2.1 High-cell-density fermentation

As previously mentioned, the cell density in a fermentation affects the biocatalyst specific activity in a whole-

cell process. The cell density can be improved by optimization of the fermentation conditions, e.g. media

optimization, or by applying different feeding strategies of the carbon source.

Here a high-cell density fermentation of Pseudomonas putida GS1 was developed by applying an exponential

feed of the carbon source (Equation 3.2) in the phase for biomass growth.

Mirata and co-workers have determined the maximum specific growth rate (μmax) of P. putida GS1 to be 1.43

h-1 in the absence of perillic acid (Mirata et al., 2009). In the first batches of fermentations, μ was set to 0.25

h-1 based on previous work from Sun and co-workers with a similar Pseudomonas putida strain (Sun et al.,

2006). However, at this growth rate oxygen deprivation occurred approximately 10 hours after the start of

exponential feed and it was not possible to keep dissolved oxygen concentration above 30% air saturation.

The growth rate was then lowered to 0.1 h-1 and the progress curve for the fermentation is shown in Figure

3.14.

74

Figure 3.14 Progress curve for a fed-batch fermentation with exponential feed keeping a constant specific growth rate of 0.1 h-1. The vertical lines represent the time points at which the exponential feed started and

finished.

By having an exponential feed of the carbon source that kept the growth rate constant at 0.1 h-1, it was

possible to reach an OD600nm of approximately 55, which corresponds to a biomass concentration of

approximately 19 g cdw/L. According to Equation 3.2, after 24 hours of exponential feeding 18.48 g cdw

should theoretically be produced in the fermenter. Therefore, it was possible to achieve the expected

biomass concentration.

It was possible to successfully increase biomass concentration from an OD600nm of approximately 25 in the

end of the base-case fermentation, which corresponds to approximately 10 g cdw/L, to an OD600nm of

approximately 55 at the end of the high-cell density fermentation, which corresponds to approximately 19 g

cdw/L.

But even though biomass concentration was considerably increased, it was not possible to verify if this leads

to a higher space-time yield of the bioprocess since (R)-perillic acid formation was not observed. Even though

(R)-limonene was added to the fermentation media in the end of the exponential feed, (R)-perillic could not

be detected in any of the HPLC chromatograms.

0

5

10

15

20

25

0

10

20

30

40

50

60

0 20 40 60 80

[bio

mas

s] (

g cd

w/L

)

OD

60

0n

m

Time (h)

OD600nm [biomass] (g cdw/L)

75

3.7 Concluding remarks

In this chapter the biocatalytic conversion of (R)-limonene to (R)-perillic acid by a wild type strain of

Pseudomonas putida was benchmarked with other possible routes for perillic acid production, namely with

the chemical synthesis starting from β-pinene developed by Wang and co-workers (Wang et al., 1993).

Reaction-related green chemistry metrics and EHS hazards of the different chemicals used in the different

routes proved to be useful tools for benchmarking of the different synthetic routes and showed that the

biocatalytic route has a greater potential to be a sustainable route for perillic acid production. However, even

though atom economy and carbon efficiency show that the biocatalytic conversion of (R)-limonene to (R)-

perillic acid by P. putida is a very promising route where the principles of maximizing the incorporation of

starting materials into the final product are obeyed, reaction mass efficiency revealed how it is still far from

reaching its full potential. It is not surprising that the reaction yield and volumetric productivity of the

reaction are still far from the values necessary for industrial implementation given the fact that the strain

that is used is a wild-type strain, which did not evolve to work in industrially relevant conditions.

The bioconversion of (R)-limonene to (R)-perillic acid by a whole-cell biocatalyst was further explored by

means of an economic and environmental assessment. A process model was developed in SuperPro

Designer® software and a sensitivity analysis on different process metrics and its influence on production

costs allowed the identification of space-time yield as the most critical process metric.

Following the identification of low space-time yield as the main bottleneck of this bioprocess, different

parameters affecting space-time yield and strategies to improve it were reviewed.

In order to understand the reasons for a low space-time yield some additional experimental work was

developed. The reaction rate was measured over time and it was seen a considerable decrease in reaction

rate approximately 8 hours after the start of the reaction. The reduction in reaction rate could be explained

by product inhibition and substrate loss by stripping of (R)-limonene in the exhaust gas, which was

demonstrated experimentally.

The strategies for improvement of space-time yield include both techniques to modify and improve the

biocatalyst and process technologies. One process technology the development of a high-cell density

fermentation, was tested experimentally in order to exemplify some of the strategies that can be applied in

the development of this bioprocess and the improvements that can be obtained with this type of tools. Even

though the cell density was successfully improved in the fermenter, the improvements in space-time yield

could not be verified. The fermentation process showed a very high instability and variability on the

76

production levels of (R)-perillic acid, which is a major concern for industrial implementation of this

bioprocess.

Several other strategies for improvement of space-time yield should be tested experimentally, including

protein and metabolic engineering for biocatalyst modification. The bioprocess analyzed in this chapter used

a wild-type strain of Pseudomonas putida. However, as reviewed in section 1.3 most processes for production

of natural products that have reached commercial scale use genetically improved microorganisms. The

genetic modification of the strain could also potentially improve the robustness of the strain, which is an

important factor for industrial implementation. On the other hand, the natural label of this product is

important for its application in cosmetic or food industry and the use of genetically modified organisms is

not well accepted by many consumers, especially in the European market. But could this bioprocess ever

reach the necessary yields and productivities for industrial implementation without the use of an improved

biocatalyst? Further experimental work on the process would be necessary to answer this question.

77

4. Case-study II: Fermentation of recombinant Pichia pastoris

for brazzein production

4.1 Introduction

In this chapter the second case-study that was explored in this thesis is introduced. It corresponds to the

production of brazzein by fermentation of the yeast Pichia pastoris. Brazzein is a natural sweetener present

in the wild African plant Pentadiplandra brazzena Baillon that is much sweeter than sucrose by several orders

of magnitude. It is a protein comprising of 53 amino acid residues, and therefore this case-study differs from

the first one since the product is a protein associated to cells’ growth. This bioprocess is classified as a

fermentation process and not a biocatalytic process.

This chapter is organized in six different sections. First a brief introduction to different sweeteners and its

productions processes is made in section 4.2. Following that, the fermentation process for brazzein

production is described in section 4.3. A very brief evaluation of the economy of the process is shown in

section 4.4 where the key process metrics of this bioprocess are analyzed. But the main focus of this case-

study is on the tools for environmental assessment and on the challenges of comparing the environmental

profile of different types of processes in different stages of development. In section 4.5 the environmental

profile of brazzein fermentation process is first benchmarked with different routes for brazzein production

(extraction from the African fruits and extraction from genetically modified maize), and secondly with

production processes of other sweeteners, namely stevia and aspartame. Finally, section 4.5.4 presents some

concluding remarks on this case-study.

4.2 Sweeteners

Sweet-tasting substances, generally designated as sweeteners, have gained increasingly importance in food

industry over the last decades. This industry has been for long in a demand to substitute sugar by a sweetener

that has the same taste and texture as sugar but with less calories. The increase in obesity and diabetes in

industrialized countries has resulted in governmental actions to promote health and wellness, e.g. Mexico

has introduced a one-peso-per-liter tax on sugar-sweetened soft drinks in 2014. In addition, consumers

78

became more concerned with their health and are increasingly demanding. Nowadays, consumers seek for

healthier products but with the same tasting and texture characteristics as sugar-sweetened products.

One of the first sweeteners to be widely applied in several food products was aspartame and it is still today

the most common intense sweetener with a worldwide consumption of 10,000 tons/year. Aspartame is a

chemically synthesized sweetener that was discovered in 1965, it is 160 to 220 times sweeter than sucrose

and its taste is described to be very similar to sucrose, showing no bitter or metallic aftertaste. It is a dipeptide

composed of two amino acids, L-aspartic acid and the methyl ester of L-phenylalanine. The synthetic route

of aspartame is represented in Figure 4.1 (Zorn and Czermark, 2014). However, aspartame safety has been

subject of several controversies and critics have linked aspartame to several health problems, including

cancer. Even though several scientific publications dismiss the potential health risks of aspartame and it is

considered as a safe food additive by worldwide health and food organizations, aspartame became a very

unpopular sweetener and consumers seek new sweeteners preferably from natural sources.

79

Phenylalanine

ProtectionMethylation

Condensation

Deprotection

Methyl L-α-aspartyl-L-phenylalaninate

Aspartic acid

Figure 4.1 Chemical synthesis of aspartame.

So as a result of the current global demand for healthy and natural products, there is a clear market for novel

sweeteners to be applied in beverages and food products.

Stevia, a natural sweetener present in high concentrations on the South American plant Stevia rebaudiana

Bertoni, has become a popular alternative to sugar and is already being used by some of the world’s biggest

beverage brands, e.g. Coca-Cola Life®. Stevia is the common designation given to steviol glycosides (shown

in Figure 4.2) that naturally occur in Stevia rebaudiana and has been consumed for centuries in Paraguay and

Brazil.

80

Figure 4.2 Structure of steviol glycosides. R: mono- or disaccharide residues.

The most abundant glycosides in the plant are stevioside and rebaudioside A, with a content of 4-8.5 % (w/w)

of stevioside and 1.5-5% (w/w) for rebaudioside A. The ratio of the different components varies depending

on the plant, and it can be manipulated by breeding aiming at an increase in Rebaudioside A, the product

with the best sensory properties. Depending on the composition, steviol glycosides are 200-300 times

sweeter than sugar. However, they have a bitter and licorice aftertaste, which is an undesirable property in

the food industry. Steviol glycosides have been approved as food additives in 2011 by the European Food

Safety Authority (European Food Safety Authority (EFSA), 2010). The manufacturing process of stevia, which

is described in more detail in section 4.5.2.3, consists of three basic steps – i) water extraction from the plant

leaves, ii) preliminary purification with ion exchange chromatography, and iii) recrystallization from methanol

or aqueous ethanol. There are several variations of this basic process and several different approaches to

extract and purify the molecules have been investigated (Puri et al., 2011). A novel process for stevia

production by fermentation of recombinant yeast has also been developed by Evolva (Mikkelsen et al., 2014)

and will start commercial manufacturing in a collaboration with Cargill®.

In this quest for the perfect sweetener with great taste and no calories, brazzein appears to be a very

promising alternative. It is much sweeter than conventional sucrose and stevia and it does not have the

undesired licorice aftertaste. Brazzein is a sweet protein from a wild African plant Pentadiplandra brazzena

Baillon. It is a single chain polypeptide, which is comprised of 53 amino acids residues. It has a molecular

weight of 6,5 kDa. The protein was discovered and named by Ding Ming and Göran Hellekant in 1994

(Hellekant and Ming, 1998). The content of brazzein in the fruit is about 0.2 - 0.05 % by weight and mainly

81

placed in the pulp. Pentadiplandra brazzeana grows in central Africa, more precisely from Nigeria east to the

Central African Republic and south to Democratic Republic of the Congo and Angola. It grows as a shrub or a

liana. The sweet fruit, and other parts of the plant, have been used for centuries and are still in use by local

people. But there is not an agricultural cultivation of the plant. The fruits are harvested from wild plants.

Brazzein is 500 times sweeter than a 10 % (w/v) sucrose solution. In comparison with a 2 % (w/v) solution,

brazzein is 2000 times sweeter. The taste of brazzein is described as purely sweet, with no sourness, saltiness

or bitterness, but the mouth feeling differs from the sweetness of sucrose. The sweetness grows slower, is

easily washed from the tongue and does not last long. Brazzein is a thermo stable protein. The sweetness

lasts even after incubation at 80 °C for 4 h.

An obvious possibility for the production of brazzein is the extraction from fruits of Pentadiplandra

brazzeana. However the large scale production of brazzein from natural sources is described as

uneconomical. The reasons indicated are limited availability of the fruit and complications associated with

large-scale production of the native plant (Lamphear et al., 2005). Therefore, the production of brazzein in a

recombinant host appears to be an interesting approach. The genetic code of brazzein has been sequenced

in 1994 by Hellekant and Ming (Hellekant and Ming, 1998) and since then there has been several attempts

to produce brazzein in different hosts. One of the options, which is further analyzed in this chapter, is the

expression of the protein in maize which has been demonstrated by Lamphear and co-workers (Lamphear et

al., 2005). One of the advantages of this option would be that the final product is sweet maize flour which

simplifies the purification. Assadi-Porter and co-workers have reported in 2000 the production of brazzein in

Escherichia coli (Assadi-Porter et al., 2000) and the company Natur Research Ingredients Inc. is currently

working on the commercialization of brazzein produced by fermentation of E. coli under the name Cweet™.

BRAIN AG has focused on the expression of brazzein in yeast cells from the species Pichia pastoris, and this

process is further analyzed throughout this chapter.

4.2.1 Intensity of sweetness

As mentioned before sweeteners have different intensities of sweetness. A high intensity sweetener, such as

brazzein, can be up to thousands times sweeter than sucrose. So the sweetness of a substance is represented

in terms of sweet potency, also called relative sweetness, which expresses how much sweeter a substance is

in comparison to sucrose. Sweet potency is correlated on a mass basis, which means that solutions with

different mass concentrations of the sweetener are compared with sucrose solutions with different

82

concentrations in a dose-response diagram. For example, brazzein has a sweetness potency of 2000 when

compared to a 2% sucrose solution, which means that brazzein is 2000 times sweeter than a 2% solution of

sucrose.

Since sweeteners have different sweetness intensities, it is important to take this into account when

comparing the environmental impacts of the different sweeteners. Therefore, a comparison in terms of

sweetness potency was made as shown in section 4.5.

4.3 Bioprocess description

The process flowsheet for the fermentation of Pichia pastoris for brazzein production is shown in Figure 4.3.

The data used for the process flowsheet and for the mass balances is experimental data from 10 L scale

experiments run at BRAIN AG facilities. The mass balances for the whole process are shown in detail in the

Appendices.

83

6 g Brazzein (98% purity)

DSP

USP

Fermentation

10.3 L

91 g/L cell dry weight

2.022 g/L Brazzein

Centrifugation

Clarification

Concentration

Buffer Exchange

Chromatography

Freeze drying

6184.1 g

6275.3 g

2059.6 g

2032 mL

1300 mL

Water 250 mL

18 L running buffer

Cultivation for 10 L:

Medium 5.5 L

Inoculum 0.5 L

Base 615.7 g

Acid 50.3 g

Antifoam 5 g

Glycerol (80%) 688.4 g

Methanol 2313 g Samples 30.6 g

Waste 45.1 g

Waste 4549.2 g

17956.8 g buffer

5900 mL buffer3200 mL running buffer

2000 mL elution buffer

Cells 3701.7 g

Figure 4.3 Process flowsheet for brazzein production by fermentation of recombinant Pichia pastoris.

84

The established bioprocess for brazzein production corresponds to a fed-batch fermentation of Pichia

pastoris, where glycerol and methanol are used as carbon sources. The media composition is shown in Table

4.1. 1 mL/L of trace metals solution, whose composition is shown in Table 4.2 is added to the fermentation

broth in the beginning of the fermentation. The fermentation is aerobic and takes place at a constant

temperature of 30 °C and at a constant pH (balanced with phosphoric acid (17% (v/v) and/or ammonia (25 %

(v/v)), and lasts for approximately 50 hours. Foam production is controlled through addition of antifoam of

polypropylene glycol. Brazzein is produced by the cells with a product yield of 0.012 g/g of methanol and

excreted to the media.

Table 4.1 Composition of the fermentation media for fermentation of Pichia pastoris.

Component Concentration (g/L)

Glycerol 40

KH2PO4 42.9

(NH4)2SO4 5

CaCl2 1

K2SO4 14.3

MgSO4 11.7

Table 4.2 Composition of the trace metals solution used in the fermentation of Pichia pastoris.

Component Concentration (mg/L)

CuSO4 2

NaI 0.08

MnSO4 3.0

Na2MoO4 0.2

H3BO4 0.02

CaSO4 0.5

CoCl2 0.5

ZnCl2 7

FeSO4 22

Biotin 0.2

H2SO4 1 μL/L

85

Since the product is extracellular, cell lysis is not necessary. After fermentation, there is a centrifugation for

separation of the cells from the fermentation broth. Following that there is a clarification step through

filtration at 4 °C and a concentration step at 10 °C by ultrafiltration where proteins and particles bigger than

50 kDa are removed from solution. Finally there is a diafiltration step for exchanging the product to a different

buffer of sodium acetate (10 mM) and sodium chloride (50 mM) and an ion exchange chromatography where

brazzein is selectively bound to the resin. The product is then eluted from the chromatography column using

a buffer of sodium acetate (10 mM) and sodium chloride (1 M). The last step is drying the product in a freeze-

drying step until a final concentration of 98% is achieved. However, the downstream processing of this

process is highly inefficient with a yield of 28.8%. Approximately 71.2% of brazzein that is produced in the

fermentation is lost in the purification steps.

4.4 Bioprocess evaluation

In the same way as for the first case-study the key process metrics described in section 2.1.1 have been

calculated and are shown in Table 4.3. In this case-study the bioprocess in analysis is a fermentation and not

a biocatalytic reaction, which means that there is not a reaction yield but a yield of product on carbon source.

This yield on carbon source should be compared with the maximum theoretical yield calculated from the

metabolic pathway. However, this comparison was not done since the pathway for brazzein production in

Pichia pastoris could not be found in literature.

Table 4.3 Process metrics of brazzein production process.

Product concentration (g/L) 2.02

Space-time yield (g/L/h) 0.075

Yield of brazzein on glycerol (g/g) 0.035

Yield of brazzein on methanol (g/g) 0.012

This process refers to a fermentation for production of a protein and not a chemical. In this case, the

requirements for the process metrics are lower since normally the selling price of proteins, which are typically

used for therapeutic purposes, is higher than the price of chemicals. However, brazzein is not a therapeutic

protein. It is aimed to be applied in the food industry, so its price should be lower than therapeutic proteins.

86

The minimum values for the process metrics that guarantee economic feasibility of the process depend on

the final selling price of brazzein.

4.5 Environmental assessment

There are only two possible routes for brazzein production, the extraction from natural sources or the

biosynthesis. The chemical synthesis is not an option since brazzein is a protein.

In this section of the thesis, an environmental comparison of different production processes was made and

is graphically summarized in Figure 4.4. In section 4.5.3 the biosynthetic route for brazzein production using

Pichia pastoris as a production host is benchmarked with two possible extraction routes – extraction from

the African fruits of Pentadiplandra brazzena and extraction from genetically modified maize. The

biosynthetic route for brazzein production is then also benchmarked with the production processes for other

two sweeteners. The two different sweeteners used as a benchmark are stevia and aspartame. The method

selected for stevia production was stevia extraction from the leaves of Stevia rebaudiana, which is the

process that is currently used for commercial production of stevia. It would also be interesting to include in

the assessment a comparison with stevia production by yeast fermentation, since this process should be

similar to Pichia pastoris fermentation for brazzein production. However, it was not possible to include it due

to lack of published data. The method selected for aspartame production was the enzymatic reaction, with

the enzyme thermolysin.

It is important to note that all these processes are in different stages of development. Stevia and aspartame

are commercially available and their production processes have gone through an extensive development and

optimization until commercial production was economically feasible. Brazzein production by fermentation of

Pichia pastoris or extraction from modified maize are still in research and development and have only been

applied at laboratory scale. Finally, the production of brazzein by extraction of the African fruits is only a

conceptual process based on the patent from Hellekant and co-workers (Hellekant and Ming, 1998).

87

Figure 4.4 Graphical summary of the different processes benchmarked in the environmental assessment.

4.5.1 Scope of the environmental assessments

In environmental assessments the first step is to define the scope of the assessment. In general, a more

holistic approach will give more meaningful results. However, the amount of data required and the

complexity of the analysis will also increase. Therefore, a compromise between the complexity and the

significance of the analysis needs to be reached.

The processes analyzed in this chapter are different types of processes – three of the five processes are

extraction from natural sources, one process is a fermentation and the last is a chemical synthesis (aspartame

production). Therefore it is not straightforward to define a common scope for the five processes. Apart from

the production of aspartame, all the other processes have in common a phase for growing of the biomass,

where the product is included. This phase corresponds to the growth of yeast cells in the fermentation

process and to the growth and cultivation of plants in the extraction processes. This phase is the upstream

processing where the product is formed. The resulting biomass containing the product is then sent to

downstream processing where the product is separated from the biomass and further purified. A scheme

representing these different steps that constitute the scope of the assessment is shown in Figure 4.5.

88

In the case of aspartame production, only the chemical synthesis of aspartame is considered in the upstream

processing. The starting point for the synthesis are the two amino acids, L-aspartic acid and D,L-

phenylalanine. The synthesis of the amino acids as well as all the other reactants is not included in the

assessment.

Figure 4.5 Scope of the environmental assessment.

The system boundaries end at the last step of the manufacturing process, the so called cradle to gate

approach. Packaging and shipping of the product are not included, neither the use nor disposal of the

product.

4.5.2 Modeling and inventory of the processes

An extensive literature was made to collect data on all the different processes. In this section all the processes

are described in detail as well as all the assumptions made, apart from brazzein fermentation which is

described in section 4.3. The mass balances of all the processes are shown in the Appendices.

4.5.2.1 Brazzein extraction from genetically modified maize

A conceptual process for brazzein extraction from genetically modified maize has been developed based on

the work from Lamphear and co-workers (Lamphear et al., 2005). One advantage of this process option would

be that after dry-milling of the seeds, the sweet maize flour could be used directly in food products. However,

89

for a meaningful comparison with the other processes for brazzein production, a process for brazzein

purification is assumed. Furthermore, the sweet maize flour could not be used in baking processes at

temperatures higher than 98 °C, which would limit the options for product formulation. In their work,

Lamphear and co-workers have proposed a process for purification of brazzein. It was developed for

analytical purposes but it is very similar to other processes for purification of recombinant proteins recovered

from plants (Menkhaus et al., 2004) (Ma and Wang, 2012). Therefore, the same conditions were assumed in

the conceptual process developed for the environmental assessment, whose flowsheet is shown in Figure

4.6.

The process starts with the cultivation of the genetically modified maize. It was assumed that the cultivation

of genetically modified maize would be done in the same conditions as regular maize, with the same yield of

maize per hectare, fertilizer and water consumption, etc. The amount of fertilizers, pesticides, manure, etc.

consumed was taken from the Agri-footprint® LCA Food Database available in the LCA software SimaPro. The

values used are for corn cob mix (CCM) production (CCM consists of the grains and cobs of maize) but it is

assumed that they are the same as for maize cultivation (Vellinga and Boer, 2012). The yield of maize per

hectare was taken from the Federal Statistical Office of Germany. The average yield was 10 ton/ha for the

years 2012-2014. The maize that is harvested needs to be matured and dried. The harvested maize is

composed of 35% (w/w) of dry mass (Federal Statistical Office of Germany, 2016). Hence 3.5 tons of dry

weight of maize is obtained per hectare.

Lamphear and co-workers measured an average brazzein content of 50 μg per gram dry weight of maize.

Single seeds reach a concentration of brazzein of 400 μg per gram of seed. This is still a process in early stages

of research, therefore the authors of the work expect an increase in brazzein content through backcrossing

of lines. Thus, a yield of 400 μg per gram of dry weight of maize was assumed (Lamphear et al., 2005).

The process for brazzein extraction starts with standard dry milling of the maize. The milling process leads to

a concentration of brazzein, since the germ of maize which is the part of maize with the highest concentration

of brazzein is separated from the hulls. The germ-rich fraction of maize is then sent to downstream

processing, which starts with an extraction step with a buffer of 20 mM of sodium acetate and 30 mM of

sodium chloride. In the extraction step the ratio solids:buffer is 5:1. After extraction, there is a centrifugation

and filtration with a 0.2 μm filter to separate the particulate matter. Then a cation exchange chromatography

separates brazzein from other proteins. Finally, there is an ultrafiltration at 3 kDa for concentration of the

solution and desalting. The final step is freeze-drying (Lamphear et al., 2005).

The method described by Lamphear and co-workers does not include the yield of each individual step. Since

the purification process is very similar to the one from the fermentation of Pichia pastoris, the same yield of

90

28.8 % was assumed for the downstream processing. The conditions in each individual step, e.g. amount of

buffer used, were also considered to be the same as in the fermentation process unless otherwise stated. It

was also assumed that the product has the same purity of 98% in the end of the purification process.

91

DSP

USP

Maize cultivation and

harvesting

Dry milling

Maize grains: 3.5 tons dry weight

Brazzein: 0.04% of maize dry weight

Extraction with buffer

(pH= 4.0, 4 °C, mixing for 1 h)

700 kg

Germ-rich fraction

Centrifugation

Filtration

Ion exchange Chromatography

Ultrafiltration

Desalting

Freeze drying

4212 kg

Brazzein: 0.4 g/L

3512 kg

Solids:buffer ratio: 1:5

Sodium Acetate 5.7 kg

NaCl 6.1 kg

Water 3500 L

Buffer:

NaCl 205 kg

Water 7024 L

403 g brazzein (98% purity)

Cultivation of 1 ha:

N-fertilizer 30 kg

Lime 400 kg

Manure 170 kg

Pesticides 0.9 kg

Maize middlings 2800 kg

Waste

Waste

Waste

Waste

700 kg

Waste

Waste

Figure 4.6 Process flowsheet for brazzein extraction from genetically modified maize.

92

4.5.2.2 Brazzein extraction from the African fruits of Pentadiplandra brazzeana Baillon

A theoretical possibility for the production of brazzein is the extraction from the African fruits of

Pentadiplandra brazzeana Baillon. The patent from Hellekant and Ming protects among other things the

isolation of brazzein from the African fruits and even though the method described is not at industrial scale

it is here assumed to be a possible method to produce brazzein (Hellekant and Ming, 1998).

A conceptual process for brazzein production through extraction from the African fruits is proposed and the

flowsheet is presented in Figure 4.7.

The production process starts with harvesting the fruits of the plant Pentadiplandra brazzeana Baillon. The

large scale cultivation of the plant is complicated and it was not possible to find data in literature for the

agricultural cultivation of this plant. Therefore, it was not included any water for irrigation of the plants,

pesticides or manure. It was just considered that the fruits are harvested from the wild. The content of

brazzein in the ripe fruit is described in the range 0.05%-0.2% by weight (Hellekant and Ming, 1998). An

average value of 0.125% (w/w) of brazzein was assumed in this environmental assessment.

Brazzein only exists in the soft pulp of the fruit, so the first step is the separation from the seeds which are

considerably large. Hence this separation would be easy to perform manually or by mechanic means. Given

the large size of the seeds, it was assumed that the mass of pulp is approximately the same as the mass of

seeds. The next step is a solid-liquid extraction using an acidic buffer (pH 5) at a ratio of solid:liquid of 1:40

(w/v). Afterwards, there is a filtration and centrifugation to separate the solids from the liquid phase that

contains the sweet protein. The proteins are then precipitated with a high salt concentration, which is

followed by a gel filtration and a cation exchange chromatography. Afterwards there is an ultrafiltration for

concentration and desalting of the solution. Finally, there is another chromatographic step, this time an anion

exchange chromatography and the last step is drying. The yield of the purification process is stated by

Hellekant and Ming as 36%. And it is assumed that the final purity is 98% as in the previous processes. As in

the previous process, process conditions similar to the process for fermentation of Pichia pastoris have been

assumed unless otherwise stated.

93

1 kg waste

USP

Harvesting of the fruits

Separation of the soft pulp from

the epicarb and seeds

2 kg fruits

Brazzein: 0.125 %, 2.5 g

Solid-Liquid extraction

1 kg fruit pulp

Filtration

Centrifugation

Salt Precipitation:

Gel Filtration

Cation Exchange

Chromatography

Ultrafiltration

43.4 kg

Brazzein: 0.06 g/L, 2.5 g

42.4 kg

205 g

62.9 g

Sodium Acetate 164 g

Dithiothreitol 0.6 g

Glycerol 2 kg

PVP 200 g

Water 40 L

Ammonium sulfate 7132 g

Sodium hydroxide 1,4 g

31.4 g

2 kg fruits

1 kg seeds

DSP

Precipitate

Salt Precipitation: Ammonium sulfate 17550 g

Precipitate 670 g

Anion Exchange

Chromatography

Drying

0.9 g Brazzein (98% purity)

NaCl 39.2 g

NaAc 2.3 g

Water 1.34 L

NaAc 0.7 g

NaCl 12 g

Water 0.41 L

Tris-HCl 0.5 g

NaCl 0.4 g

Water 63 mL

Figure 4.7 Process flowsheet for brazzein extraction from the African fruits of Pentadiplandra brazzeana.

94

4.5.2.3 Stevia extraction from leaves

Stevia is commercially produced by extraction of the glycosides from the leaves of the plant Stevia

rebaudiana Bertoni. The process used for the environmental assessment is based on the patent from the

company PureCircle that commercializes Stevia (Purkayastha et al., 2012). The process described in the

patent fulfills the European requirements to extract stevioside and rebaudioside A with high purity. The

process consists of the following main steps – brazzein is extracted with hot water from the dried and

powered leaves, then there is a pre-treatment with base and iron chloride, filtration and ion exchange

chromatography and the purification consists of several filtration, drying and crystallization steps.

The process flowsheet is sub-divided into two parts which are shown in Figure 4.8.

In the environmental assessment it is also considered the cultivation of the plants. The data for the cultivation

of the plants is taken from a report on stevia cultivation from the Rural Industries Research and Development

Corporation (RIRDC) and the Australian Government (Midmore and Rank, 2002). The crop yield is in average

1.6-2 tons dried stevia leaves/ha/year. The average value of 1.8 tons/ha was used in this assessment. In the

cultivation of one hectare of stevia, 40 kg of nitrogen, 20 kg of phosphorus and 30 kg of potassium are used

as fertilizers. Supplementary irrigation is generally assumed to be essential to avoid any water stress on plants

unless the growing area has reliable rainfall throughout most of the year. However, it is dependent on the

amount of rainfall on the cultivation site. Therefore, water for irrigation as well as herbicides and pesticides

were not considered in this analysis, since they are site-dependent.

The plant leaves contain in average 9.1% of stevioside and 3.8% of rebaudioside A (Purkayastha et al., 2012).

After harvesting the leaves are dried, separated from the stems and powdered before extraction of the

glycosides with hot water. Afterwards, the leaf residues are removed by filtration. The following basification

of the solution to a pH of 10 should flocculate unwanted substances, which are removed by filtration. The

following step is an ion exchange chromatography that separates the glycosides from remaining substances.

Afterwards, a concentration and drying steps originate a yellow sweet powder, which is designated as

“powder 1”. This powder is basically a mixture of different steviol glycosides. This powder is then solvated in

methanol and filtered. The filtrate contains rebaudioside A and the precipitate contains stevioside. For

further purification of the stevioside the precipitate is dried and crystallized with ethanol. A series of

filtration, crystallizations and drying steps are taken until stevioside crystals are recovered with a purity of

99.3% and rebaudioside A crystals with a purity of 98.9%. The two glycosides are then mixed in the final

product which contains 29% rebaudioside A and 71% stevioside.

95

DSP

UPS: Cultivation of 1 ha

Powdering the leaves

Stripping the leaves from the stems

1 800 kg

Filtration

Solid-Liquid extraction

19 800 kg12,9 g/L of stevia

Water (pH=6,5) 18 000 kgWater to leaves ratio 10:1 (w/w)

NeutralizationpH=7, room temperature

Basification pH=10, 50° C

Calcium Hydroxide43.2 kg

FeCl395.4 kg

Celite Filtration

Filtration

Concentration and Spray-drying

Deionization and Decolourization

Ion exchange chromatography

Precipitate

Yellow powder of steviol glycosides mixture (powder 1)

219.6 kg

64.6% Stevioside. 25.3% Rebaudisoide A.

Drying

1 800 kg9,1 % Stevioside: 163,8 kg

3,8 % Rebaudisodie A: 68,4 kg

Harvesting Stevia rebaudiana

Fertiliser : N 40 kgP 20 kgK 30 kg

96

Solvation in Methanol25° C

Drying

Filtration

134.4 kg90.6% Stevioside

Precipitate

Methanol 867.4 kg5:1 v/w

Crystallisation with Ethanol10-12° C

Vaccum drying

Filtration

Filtrate

268.8 kg Ethanol (90%)

99.3% Stevioside

129.1 kg

Evaporation of the Alcohol

Ultrafiltration (2.5 kDa)

Dilution with water

Concentration and Spray drying

89.6 kg60% Rebaudioside A

Solvation in Ethanol (50° C)

Drying

Filtration

67.2 kg89,8% Rebaudioside A

Precipitate

Crystallisation with Ethanol12° C, slow agitation

Drying

Filtration

Precipitate

98.9% Rebaudioside A

51.8 kg

96,2% Ethanol 448 kg5:1 v/w

92% Ethanol 134.4 kg

Water 896 kg

Precipitate

Final product:180. 9 kg71% Stevioside 29% Rebaudioside A

Powder 1219.6 kg

Filtrate

134.4 kg90.6 % Stevioside

Figure 4.8 Process extraction for stevia extraction from the leaves of Stevia rebaudiana Bertoni.

97

4.5.2.4 Aspartame enzymatic production

There are different methods for production of aspartame. The condensation of the two modified amino acids,

phenylalanine and aspartic acid, can be a chemical or an enzymatic reaction (Hanzawa, 2010). In this case an

enzymatic production of aspartame was selected.

The data for the mass balances of this process is taken from a patent from the company Ajinomoto Co., Inc.,

which is one of the major producers of aspartame, and from a synthesis procedure from Lindeberg (Kiyotaka

Oyama et al., 1985) (Lindeberg, 1987).

The process flowsheet for the enzymatic production of aspartame is shown in Figure 4.9.

The process starts with the preparation of the two amino acids. L-aspartic acid needs a protection group for

the amino group - a Z-protection. So a N-benzyloxycarbonyl group is introduced in alkaline media, to form N-

benzyloxycarbonyl-aspartic acid (ZA). Afterwards, the protected amino acid is washed with an organic

solvent, toluene, to remove unreacted compounds. DL-phenylalanine is prepared through esterification in

methanol in acid conditions, to form phenylalanine methyl ester (PM). Afterwards, methanol is evaporated

through distillation and the amino acid is dissolved in water.

The aqueous solutions of the two protected amino acids are mixed with the enzyme thermolysin and the

condensation reaction takes place at 40 °C. The L- form of the methylated phenylalanine, L-phenylalanine

methyl ester (L-PheOMe) reacts stereo and regio selectively with the N-protected aspartic acid. The reaction

product, Z-Aspartame, forms an insoluble addition compound with one molecule of remaining D-PheOMe

and precipitates. When the reaction is completed, the Z-Aspartame-D-PheOMe addition compound is

collected as a solid product and thermolysin is recovered from the solution. D-PheOMe is then separated

from Z-Aspartame by acid washing with hydrochloric acid. The protection group, Z- is removed from Z-

aspartame by hydrogenolysis. Finally, aspartame, or α-L-aspartyl-L-phenylalanine methyl ester, is obtained

by crystallization and drying. D-PheOMe is recycled after racemization by alkaline treatment and

reesterification in methanol (Hanzawa, 2010).

For simplification of the mass balances, the reuse of the enzyme and the Pd/C catalyst used in the

hydrogenolysis step was not considered as well as the recovery of D-phenylalanine methyl ester. It was

assumed a loss of 10% of product in the purification steps.

98

USP

Condensation (40 °C)

Filtration

Acidification

Z-Aspartame-D-PheOMe 0.1712 mol

Hydrogenolysis

Crystallization

Drying

Aspartame

45 g

DSP

Protection reaction (0-12 °C,

pH= 9.5-12)

L-aspartic acid

29 g

105 g Toluene

Methanol 301.8 g

HCl 39.8 g

NaOH 11.3 g

Water 113.1 g

Esterification reaction

Water 136.4 g

Aqueous PM solution

256 g

DL-Phenylalanine

90.1 g

Phase separation Distillation

Thermolysin 7.2 g

Water 265 g

Calicum acetate monohydrate 1.3 g

29.5 g NaOH

43.6 g Benzyloxycarbonyl chloride

106 g Water

Aqueous ZA solution

198.7 g

Water 510 g

Ammonium formate 21.4 g

Water 4800 g

Catalyst Pd/C (50%) 8.5 g

Methanol 403 g

D-PheOMe (recycle…)

Water 170 g

HCl 6.2 g

Toluene phase Methanol

Figure 4.9 Process flowsheet for aspartame production process.

99

4.5.3 Benchmarking of brazzein fermentation process with similar processes

4.5.3.1 Process Mass Intensity

Since PMI is one of the environmental metrics more widely used, it was calculated for the five production

processes previously described and it is shown in Table 4.4.

Table 4.4 Process mass intensity for the five different processes analyzed in the environmental assessment.

Brazzein

fermentation

Brazzein extraction

from maize

Brazzein extraction from the

African fruits

Stevia production

Aspartame production

Total PMI (kg materials/kg product)

5,647 28,130 76,576 115 155

The PMI values shown in Table 4.4 include water, which is the biggest contributor to total PMI. Water

intensity of the different processes will be further discussed in section 4.5.3.3.

From the five different production processes, brazzein extraction from the African fruits of Pentadiplandra

brazzeana Baillon is the one that shows the highest PMI, approximately 77 kg of materials are required to

produce 1 gram of brazzein. This demonstrates the fact that brazzein extraction from the fruits is not an

economically or environmentally feasible process, as it was expected. Furthermore, the process for brazzein

extraction from the fruits is only a conceptual process, where all the data used in the assessment comes from

the patent describing the analytical procedure to isolate brazzein. This analytical procedure did not go

through any development or optimization, so it was expected to have a very high PMI for this process.

The second highest PMI is for brazzein extraction from maize, with a total PMI of approximately 28 kg/g. This

process is at laboratory scale like brazzein fermentation, so one can say that they are in comparable stages

of development. But the fermentation process for brazzein production shows a total PMI 5 times lower than

brazzein extraction from maize. The fermentation for brazzein production shows a total PMI of 5.6 kg/g,

which is high as it would be expected for a process that is still under development at laboratory scale. At this

stage of development, the optimization of materials consumption is not a concern as it is when the process

is at production scale. However, it also shows how the process of brazzein fermentation still requires some

work on process development and optimization. It is not uncommon that fermentation processes for

100

chemicals production at commercial scale have a total PMI below 100 kg materials/kg product. Some

examples are citric acid production, with a total PMI of 17 kg/kg, pyruvic acid production, with a PMI of 53-

35 kg/kg depending on the operations used in the downstream processing, or riboflavin with a total PMI of

39 kg/kg. But fermentation processes for recombinant proteins production show a much higher PMI, since

the downstream processing is much more complex and consists of several units of operation. One good

example is insulin , which shows a PMI of 55 kg/g (Heinzle et al., 2008).

The relation between total PMI and the stage of development of the process is again verified if one looks at

the values for PMI of stevia and aspartame production, which show a PMI of 120 kg/kg and 160 kg/kg

respectively. These two values are higher than the above mentioned values for fermentation processes for

chemicals production but are in any way 50 times lower than the fermentation process for brazzein

production. So there is a clear relationship between the development stage of the process and total PMI,

which would be expected since processes at commercial scale have gone through an intensive research and

development. Clearly it is not appropriate to compare processes in different development stages, they will

obviously show values very different for PMI as well as all the other metrics. But on the other hand, how else

can the benchmarking of novel natural products be made? The benchmarking of novel products with

products that are already at commercial scale can provide a development target for processes that are still

under research and development, like brazzein fermentation.

It is also interesting to put these values into context with sucrose production. For processing of sucrose, total

PMI is 10-25 kg/kg if water is included and 0.4 kg/kg excluding water (Gerbens-Leenes and Hoekstra, 2009).

These values exclude the cultivation of the biomass that is used as a source of sucrose, e.g. sugarcane or

sugar beets.

But perhaps a more interesting analysis is to see how much can PMI of the fermentation process decrease if

the process is improved. Therefore, a sensitivity analysis on PMI was performed to analyze how much it is

affected by the product concentration and the yield of downstream processing and the results are shown in

Figure 4.10.

101

Figure 4.10 Variation of PMI of brazzein fermentation process for different product concentrations and different yields of downstream processing.

It is clear that PMI is very much affected by product concentration in the fermentation broth and by the

efficiency of DSP. If during the process development and optimization phase product concentration is

increased to 10 g/L and the efficiency of downstream processing is improved to 80%, PMI would be 10 times

lower, approximately 400 kg/kg. Even though this is a considerable improvement of the process, it is still not

enough to reach PMI values comparable to the commercial processes for stevia or aspartame production, so

further optimization would be required. If product concentration could be improved to approximately 35 g/L

and downstream processing efficiency kept at 80%, PMI would be 115 kg/kg, which is a comparable value to

stevia production.

Hence, one can conclude that both upstream and downstream have a big impact on PMI and it is not enough

to improve only one of the parts of the process. Both USP and DSP need to be improved to reach the full

potential of the process. The application of a sensitivity analysis proved to be a valuable tool to define targets

for process development.

However, the PMI values previously shown do not take into account the relative sweetness of the different

sweeteners. Brazzein, stevia and aspartame have different sweetness intensities and are much sweeter than

102

sucrose. Therefore, to have the same sweetness intensity in the final food products, the necessary amount

of for instance brazzein will be different from the required amount of aspartame or sucrose. Hence, a

correction of PMI taking into account the relative sweetness of the different sweeteners was done. The

sweetness potency of the different sweeteners in comparison to a 2% and a 10% solution of sucrose is shown

in Table 4.5.

Table 4.5 Comparison of the sweetness potency of aspartame, brazzein and stevia components (rebaudioside A and stevioside) in relation to sucrose. (SE – Sucrose Equivalent)

Sweetener Sweetness potency in

relation to a 2% solution of sucrose (2% SE)

Sweetness potency in relation to a 10% solution

of sucrose (10% SE) Reference

Aspartame 250 107 (Walters et al., 1991)

Brazzein 2000 500 (Hellekant and Ming, 1998)

Rebaudioside A 320 170 (Mitchell, 2006)

Stevioside 213 190 (Mitchell, 2006)

Process mass intensity of the different processes corrected on the different sweetness intensities is shown

in Table 4.6.

Table 4.6 Process mass intensity corrected on sweetness potency for 2% SE and 10% SE (SE – Sucrose Equivalent).

Brazzein

fermentation

Brazzein extraction

from maize

Brazzein extraction from the

African fruits

Stevia production

Aspartame production

PMI (kg/kg) per sweetness potency in relation to 2% SE

3 14 38 0.5 0.6

PMI (kg/kg) per sweetness potency in relation to 10% SE

11 56 153 0.6 1.4

Taking into account the relative sweetness of the different sweeteners, PMI values are considerably lower.

However, PMI for brazzein fermentation is still higher than PMI for stevia and aspartame production. On the

other hand, PMI values for brazzein fermentation (3 kg/kg per sweetness potency (for 2%SE) and 11 kg/kg

103

per sweetness potency (for 10% SE)) are now on the same range as PMI values for sucrose processing (10-25

kg/kg). Taking into account that the process is still under development, once the brazzein fermentation

process reaches its full potential PMI values should be even lower. This shows that the fermentation of Pichia

pastoris for brazzein production can be an environmentally promising option for sweeteners production if it

is successfully intensified.

4.5.3.2 Product concentration

The product concentration of the different sweeteners on biomass is shown in Table 4.7, as well as the

product concentration before DSP.

Table 4.7 Product concentration on biomass in % (w/w) and in solution before downstream processing in g/L for the different production processes.

Brazzein

fermentation

Brazzein extraction from maize

Brazzein extraction from

the African fruits

Stevia production

Aspartame production

Product concentration in

biomass (% (w/w) 2.2 0.04 0.125 12.9 na

Product concentration before

DSP (g/L) 2.02 0.4 0.06 12.9 40

Looking at the product concentration in biomass and in solution before downstream processing it is clear

that there is a correlation between product concentration and total PMI. The processes with a lower product

concentration in biomass and in the solution before DSP, namely brazzein extraction from maize and from

the African fruits, are the ones with a higher PMI. On the other hand, the processes with a higher product

concentration show a lower PMI. This again shows the importance of product concentration for both the

economy and the environmental profile of a process.

It is also interesting to note that stevia leaves have a quite high concentration of sweetener, which grants a

natural advantage to the process for stevia production by extraction from natural sources.

104

4.5.3.3 Water Intensity

In Table 4.8 is shown the water intensity and the solvent intensity of the five different processes.

Table 4.8 Water intensity and solvent intensity of the different processes.

Brazzein

fermentation

Brazzein extraction from

maize

Brazzein extraction from

the African fruits

Stevia production

Aspartame production

WI (kg/kg) 4,932 26,101 46,460 105 130

Contribution to total PMI (%)

87 93 61 91 84

SI (kg/kg) na na na 9.2 18.3

Contribution to total PMI (%)

na na na 8 11.5

In all the five processes water represents 60-90% (w/w) of all the materials used and is therefore the biggest

contributor to total PMI.

Solvent intensity, on the other hand is not significant. The three processes for brazzein production do not

involve the use of any solvent besides water. Stevia production has a solvent intensity of 9.2 kg solvents/kg

product. The solvents used are ethanol (4.4 kg EtOH/kg product) and methanol (4.8 kg MeOH/kg product). In

aspartame production process 18.3 kg solvents/kg product are used. The solvents used are methanol (16 kg

MeOH/kg product) and toluene (2.3 kg toluene/kg product). None of the solvents used are highly hazardous.

Toluene is classified by McElroy and co-workers as a problematic solvent (McElroy et al., 2015), but it is not

used in high amounts (it represents 1.5% of total PMI).

The results clearly show that water contributes in all of these bio-based processes with more than 50% to

total PMI. Hence, water consumption should not be underestimated and it should be included in the

calculation of PMI. These results highlight the importance of water consumption on bioprocesses. In a world

where water scarcity is increasingly problematic, there should be more research focus on water reutilization

or reduction/optimization of water consumption in biochemical processes. A bio-based industry can have

several environmental benefits in comparison to conventional petrochemical based industry. However, one

cannot forget that these bio-based processes have an inherent high consumption of water, in most cases

105

higher than in conventional chemical processes. If water use is not addressed as a topic of high priority, the

bio-based economy could potentially have a negative environmental impact in terms of water consumption.

4.5.3.4 E-factor

E-factor as defined by Sheldon, i.e. excluding water, was also calculated and is shown in Table 4.9.

Table 4.9 E-factor for the different processes.

Brazzein

fermentation

Brazzein extraction from maize

Brazzein extraction from the

African fruits

Stevia production

Aspartame production

E-factor (kg waste/kg product)

716 2,028 30,115 11 24

All of the processes for brazzein production show a high value for E-factor, as it would be expected given the

early stage of development of these processes.

E-factors of stevia and aspartame production processes are in the range of typical E-factors for the fine

chemical industry, with an annual production of 102-104 tons/year (Sheldon, 2011).

It is interesting to compare the E-factor for brazzein fermentation with typical E-factors of fermentation

processes. However, very few have been reported. Sheldon reports an E-factor of 1.4 kg/kg for the bulk

fermentation of citric acid. E-factors for the fermentation processes for the production of therapeutic

proteins (biopharmaceuticals) are much higher. The production of recombinant human insulin has an E-factor

of 6000 kg/kg, and this value increases to 50,000 kg/kg if water is included in the calculation (Sheldon, 2011).

4.5.3.5 Origin of the raw materials

It is also interesting to look at the origin of the raw materials used in the processes. The raw materials have

been grouped according to their origin and this categorization is shown in Table 4.10.

106

Table 4.10 Categorization of the origin of the raw materials used in the different processes.

Category Raw materials

Renewables Ethanol; glycerol, manure

Mining

P-fertilizer; K-fertilizer; calcium hydroxide (Ca(OH)2); Iron (III) chloride (FeCl3); sodium chloride (NaCl); phosphoric acid (H3PO4); monopotassium phosphate (KH2PO4); calcium chloride (CaCl2); potassium sulfate (K2SO4); magnesium sulfate (MgSO4); lime (CaCO3); sodium hydroxide (NaOH); catalyst Pd/C

Petrochemical Methanol; N-fertilizer; ethanol; ammonia; sodium acetate; antifoam; polyvinylpyrrolidone; ammonium sulfate; benzyloxycarbonyl chloride; toluene; hydrogen chloride; hydrochloric acid; ammonium formate

Complex substances Pesticides; L-aspartic acid; DL-phenylalanine; thermolysin

The raw materials have been grouped in four different categories: renewables, which include all the natural

resources that can renew themselves naturally over time or all the materials that are produced from

renewable resources; mining, which include all the materials that are extracted from ores through mining or

derive from products extracted from ores; petrochemical are all the materials that are produced from oil or

gas; and finally complex substances are built up from several different materials. Water was not included in

this categorization of the raw materials. Since the production of biomass is included in the scope of the

assessment, stevia leaves, yeast cell or the African fruits of Pentadiplandra brazzeana Baillon are not included

as raw materials. Ethanol can be produced by conventional means from petrochemical materials or by

fermentation. Therefore, it was included in both categories and in the processes where ethanol is used it was

considered that 50% was from renewable sources and 50% from petrochemical sources.

107

Figure 4.11 Distribution of origin of raw materials of the different processes.

108

It is interesting to see in Figure 4.11 that even though all these processes are bio-based, the majority of raw

materials used are of petrochemical origin. But the fermentation process is the one with a highest percentage

of renewable raw materials (29%), mainly due to glycerol that is a by-product of biodiesel production. The

high percentage of petrochemicals in the fermentation process comes mainly from the use of methanol as

carbon source.

4.5.4 Environmental, health and safety aspects of P. pastoris fermentation

In addition to the comparison with other production processes, it was also made an identification of the

environmental hotspots of the fermentation process with the aim to draw guidelines for the improvement

of the environmental profile. Some recommendations regarding the health, safety and environmental

aspects of the process are given in this section.

The procedure that was adopted in (R)-perillic acid case-study for assessment of the health, safety and

environmental aspects of the reactants and solvents used, was also applied in this second case-study. The

flags system adopted by McElroy and co-workers (McElroy et al., 2015) and based on the H-statements of

the different chemicals was applied and is shown in Table 4.11.

Table 4.11 Analysis of the health, safety and environmental hazards of the reactants and solvents used in the fermentation of Pichia pastoris for brazzein production.

Green flag Yellow flag Red flag

Solvents Water X

Reactants Methanol H301: Toxic if swallowed H311: Toxic in contact with skin H331: Toxic if inhaled H370: Causes damage to organs

X X

The only solvents used in this fermentation process are water-based buffers which gives to this process a

“green flag” in terms of solvents use.

From all the chemicals used in the fermentation media, methanol is the only one that shows some concerning

H-sentences, namely the H-sentences related to toxicity (H301, H311 and H331) which give a “yellow flag” to

109

the use of this chemical, and the H-sentence H370 which concerns damage to organs and gives a “red flag”

to this chemical. It could be interesting to consider substituting methanol by another carbon source or to use

only glycerol as a main carbon source, if cell growth or protein expression are not affected. This substitution

would avoid using a toxic chemical and would also considerably decrease the contribution of petrochemical

raw materials to the process.

It was also verified if any of the chemicals used in the process are in the SIN List and classified as chemicals

of very high concern. Boric acid (BH3O3) has been classified as carcinogenic, mutagenic and reprotoxic (CMR)

according to Annex VI of Regulation 1272/2008 and is classified as a chemical of very high concern. It is only

used in concentrations of nanograms per liter, since it is only added as a trace element to the fermentation.

Therefore it should not constitute any danger for the safety of the workers.

4.6 Concluding remarks

The five different processes were compared in terms of PMI, water and solvent intensity, E-factor and origin

of the raw materials. Benchmarking the fermentation of Pichia pastoris with the other processes for brazzein

production (extraction from genetically modified maize and extraction from the African fruits of

Pentadiplandra brazzeana Baillon) it is clear that the fermentation process is the most promising option.

However when comparing it with the processes for production of stevia or aspartame at a commercial scale,

the fermentation process shows a much higher PMI, E-factor and water intensity which would be expected

given the early stage of development. In anyway, the benchmarking of the fermentation process with other

processes revealed itself as a useful tool since it provided some targets for process development and

optimization. Running a sensitivity analysis on the process mass intensity of the fermentation process, it was

concluded that product concentration in the fermentation broth needs to be increased to 35 g/L and the

efficiency of DSP to 80% in order to have a PMI comparable to commercial processes.

Looking at the origin of the raw materials and the EHS aspects of the chemicals used, some suggestions for

improvement of the environmental profile of the process could also be drawn. Methanol is a toxic chemical

and contributes largely to the high proportion of petrochemicals used in the process (58%). Therefore,

substituting it by another carbon source such as glycerol could be an interesting option. On the other hand,

glycerol can be produced from propylene or can be a waste product from biodiesel industry. From an

environmental point of view it would be preferable to use a raw material that is a waste from another

110

industry. The best practice is to make a comparison of different sources of the raw materials and choose the

one that is preferable from an environmental point of view. This is however a laborious task.

Some of the environmental metrics that were used in this case-study are different from the (R)-perillic acid

case-study. Since the process is a fermentation and not a biocatalytic reaction it makes no sense calculating

atom economy, reaction mass efficiency or any of the reaction-related metrics. In this case there is not a

single reaction for product formation. The product is a protein that is built from several molecules in the cell.

Developing a set of metrics specific for the environmental assessment of fermentations would be very

interesting and a valuable tool, but it is also quite a challenge given the complexity of all the reactions

involved.

It is very important to note that the quality of data is critical for an environmental assessment. In this

assessment it was necessary to make some assumptions for the modelling and inventory of the production

processes due to lack of data, and the influence of these assumptions on the final results should be analyzed

in an uncertainty analysis. Furthermore, the fermentation of Pichia pastoris for brazzein production is still at

laboratory scale and further development and optimization will still be performed. This final process

optimization, through the adjustment of the operation conditions and equipment sizes, has also an influence

in the environmental performance (Brunet et al., 2012).

Since assumptions and estimations need to be made in these environmental assessments, the accuracy of

the method is limited to a certain degree. These methods for environmental assessment should be seen as a

valuable supportive tool for process development and not as a tool to accurately measure waste production

or other environmental-related topics.

One could also argue that an economic assessment would be sufficient and there is no need to perform

environmental assessments in early stages of development, given the fact that if the economy of the process

is improved so is also the environmental profile, e.g. by improvement of PMI. However, the work done in this

case-study revealed how an environmental assessment can bring a new perspective and different

information relevant for process development, e.g. information on substitution of chemicals.

Likewise, one could argue that if we just apply environmental commonplaces such as reducing water

consumption, avoiding the use of toxic chemicals, etc. the environmental profile would anyway be improved

without having to do a full environmental assessment involving extensive literature review for collection of

data. But the environmental assessment allowed a comparison with other similar processes and allows the

research teams working in R&D to see how far their processes are from commercial processes which can help

them to define targets for process development and optimization. Obviously, the ideal process would have

111

an E-factor of zero and a PMI of 1. The goal is always to reduce E-factor and PMI, but how much can they

realistically be reduced? Benchmarking processes in early stages of development, such as brazzein

fermentation process, puts some realistic targets for E-factor, PMI and other metrics. In this case, for

instance, a realistic target for brazzein fermentation process would be a PMI in the range of 100 kg/kg. This

is a very valuable tool for scientists, e.g microbiologists or molecular biologists, working in protein and

metabolic engineering to improve the expression levels of the proteins in the cells. It puts the values for E-

factor and other metrics into context, and it would otherwise be difficult to find a meaning for them.

112

113

Part III

Discussion, Conclusion and Perspectives

114

115

5. Discussion

This chapter aims to connect the different parts of the thesis, the different tools for bioprocess evaluation,

the two different case-studies and the different strategies for bioprocess improvement. General guidelines

and recommendations for the successful implementation of processes for production of natural products are

presented.

5.1 Methodology for development of processes for the biosynthesis of natural

products

Natural products show an incredible potential for use in several industries - pharmaceutical, cosmetics, food,

etc. However, its industrial production is hindered by the difficulties in synthesizing these complex molecules.

Bioprocesses, both biocatalytic or fermentation processes, appear as a promising possibility for the

biosynthesis of these molecules and there are some examples of natural products that have been successfully

developed and applied in industrial scale, e.g. artemisin production. Metabolic engineering and protein

engineering are areas where great development is being achieved and offer great potential for the

biosynthesis of natural products.

However, the development of these processes is rather complex and requires a holistic view of the several

limitations and challenges of these processes, from a production host cell and/or biocatalyst, reaction and

process perspectives.

In this thesis two fundamentally different case-studies were assessed. The first case-study corresponds to a

biocatalytic multi-step reaction for the production of a small molecule, (R)-perillic acid which intends to be

used in cosmetic industry and could also have potential uses in pharmaceutical or food industry. The second

case-study corresponds to a fermentation for recombinant protein production. The recombinant protein,

brazzein is intended to be used in food industry. Therefore, this thesis offers a comprehensive tool for the

development of natural products by covering different types of bioprocesses, different types of natural

products and different sectors of chemical industry. As a result of the work developed for both case-studies,

a general methodology for the development of bioprocesses for natural products production was developed

and it is shown in Figure 5.1.

116

This methodology is sub-divided into three parts. The first part corresponds to a route selection/process

screening of the different options for production of a given natural product. It is generally assumed that

extraction from natural sources is not an economically feasible or sustainable alternative. However, that is

not always the case and stevia is a very good example of a process where extraction is feasible. The economic

feasibility of extraction is dependent on the amount of natural product existent in the natural source. In this

route selection all the possible routes for production of a given natural product must be reviewed and there

might be several possibilities for chemical synthesis or biosynthesis using different starting materials. This

literature review is followed by the mass balances and an environmental assessment of all the possible

options. In the environmental assessment, an analysis of the EHS hazards of all the chemicals used in the

chemical inventory of each option is required as well as the calculation of some key green chemistry metrics.

The green chemistry metrics are however dependent on the type of molecule being produced. If the product

is a small molecule produced in a single or multi-step reaction catalyzed by a chemical catalyst or a

biocatalyst, as it is the case for (R)-perillic acid, reaction-related metrics (AE, RME and CE) should be used for

comparison of the two synthetic routes. But these metrics do not make sense for assessment of large

molecules that are built up in several steps of a metabolic pathway, e.g. brazzein. In this case and also when

analyzing the extraction from natural sources, process-related metrics such as PMI or E-factor should be

preferred.

In case route selection shows that the biosynthetic route is the most promising one, an in silico evaluation of

this route should be done (part 2 of the methodology) which leads to the identification of the process

bottlenecks and strategies for its improvement. These strategies, both from a process/reaction and

biocatalyst or host cell perspectives, should then be tested and further explored in the laboratory (part 3 of

the methodology). The laboratory work is an essential part of the development of a bioprocess and it should

not be underestimated or undervalued. But an in silico evaluation of the bioprocess prior to experimental

work can guide and focus the experiments on the most critical bottlenecks.

The novelty of this thesis lies on this systematic procedure for route selection and process screening, as well

as putting the biosynthesis of natural products into context in relation to demands of an industrially feasible

process. There are several novel bioactive natural products with very interesting properties being discovered.

But the path to commercialization is rather long and there is not a straightforward procedure for process

design and process development of these novel products. Developing novel natural products until

commercialization is a challenge faced by many small-medium sized biotechnology enterprises and this thesis

tries to understand how these bioprocesses can be developed and offer some guidelines for process

development.

117

Moreover, it reports the environmental profile of bioprocesses for natural products production, which is not

often found in scientific literature. The early-stage environmental assessment that was applied in both case-

studies can give very meaningful insight into process development.

Besides the environmental assessment, different tools for exploring the economy of a process are described

throughout the thesis. Calculating process metrics, such as product concentration, space-time yield,

biocatalyst yield or reaction yield, is a simple tool that does not require a big amount of data. But it can give

very meaningful information on the process, which can guide research and considerably reduce development

time. Hence, it is recommended to calculate process metrics at the start and throughout process research

and development.

But setting targets for the process metrics for product and process development can be quite challenging,

especially for novel products, such as (R)-perillic acid and brazzein. The demands on process metrics are

dependent on the product annual production and price. Bulk chemicals, with an annual production in the

range of 104-106 ton/year and a very low selling price (in the range of 1 €/kg) have much tighter demands on

process metrics than for instance pharmaceutical chemicals with an annual production of 10-103 ton/year

and a selling price that can be higher than 100 €/kg. With a selling price as low as 1 €/kg the profit margin is

very low, therefore the process needs to be fully optimized in terms of reaction yield, product concentration,

space-time yield or volumetric productivity and/or biocatalyst yield. But for these new natural products it

can be a challenge to define a reasonable annual production and selling price. Since there are no alternative

processes commercially implemented and competing with these products, there is more freedom to set the

selling prices. But this freedom is always limited by the market where the product is to be inserted. In the

case of these two products, it would be expectable that the fact that they are produced by biotechnological

means gives some advantage for product marketing and influence positively the selling price. But overall it

can be a challenge to define targets for process research and development when there are so many variables

influencing annual production and selling price. This thesis proposes a methodology that can help setting a

price for novel natural products and guide process development for novel products without a fixed selling

price or identical competitors in the market.

118

Figure 5.1 General methodology for the development of biosynthetic processes for natural products production.

119

Scientific work on the development of novel biosynthetic routes for natural products has been widely

reported and advances on synthetic biology promise very exciting developments in this field. However, if the

commercial implementation of these bioprocesses is intended, the economic demands of an industrial

chemical process should be kept in mind throughout the developments stages. This thesis attempts to show

how the successful implementation of a bioprocess for production of natural products is dependent on a

fundamental knowledge of the biocatalyst or production host cells, reaction and process constraints that

strongly influence the process viability. The application of an in silico process analysis can be of great benefit

in guiding experimental work and suggesting guidelines for improvement of the bioprocess aiming at an

industrial scale implementation of the process. The proposed systematic assessment can support the scale-

up procedure and also shorten the development time by the early identification of potential bottlenecks and

limitations (Heinzle and Hungerbuhler, 1997).

The development of bioprocesses for biosynthesis of natural products is a complex task, with several

different bottlenecks which in most of the times requires improvements both from the biocatalyst and/or

host cell and the process. Therefore, different areas of expertise are required (microbiology, protein

engineering, metabolic and genetic engineering, process and chemical engineering, etc.) and the teams

working in the development of these bioprocesses should be multi-disciplinary. Communication can be quite

a challenge but looking at these processes from different perspectives can be the key to success.

5.2 Case-study I: Bioconversion of (R)-limonene to (R)-perillic acid by a whole-

cell biocatalyst

The first case-study was the most explored in this thesis, where the whole methodology presented in Figure

5.1 was applied. In first place, the biosynthetic route for (R)-perillic acid production was benchmarked with

a chemical synthetic route, which showed that the biocatalytic route is the most interesting one, both from

an economic and environmental perspectives, to be further explored. Following this route selection, the

bioconversion of (R)-limonene to (R)-perillic acid by a whole-cell biocatalyst was evaluated in an economic

assessment, where key process metrics were calculated, a process model was built and a sensitivity analysis

on some of the process metrics was run, which allowed the identification of the main bottlenecks of the

process and different strategies for its improvement.

Calculating the key process metrics proved to be a useful tool, but it only gave information on how far the

process still is from economic feasibility and industrial implementation. It did not give any insight or

120

guidelines on process development, meaning that it failed to rank and prioritize the different metrics. Process

modelling and sensitivity analysis of the key process metrics was a very useful tool to further explore the

different metrics and prioritize them.

The results emphasized that space-time yield is a metric of high relevance in whole-cell processes. This metric

is very complex and it is influenced by several different parameters, which can be improved in many different

ways. Whole-cell biocatalytic processes are extremely complex systems, where several different parameters

are interconnected and correlated. And addressing one parameter can have positive effects in one bottleneck

but it can actually aggravate others.

The overall methodology followed was a valuable tool to identify targets for improvement of the biocatalytic

process, which should result in a reduced process development time. Improving the economy of the process

should also be translated in terms of environmental benefits, since economic and environmental

sustainability are interconnected.

Unfortunately, it was not possible to experimentally quantify any improvement on space-time yield due to a

lack of stability of the system. In several fermentation batches no (R)-perillic acid production was observed,

which revealed a lack of consistency of the strain used. Screening of a different host should also be

investigated, although Pseudomonas sp. has an inherent competitive advantage which is its natural tolerance

to solvents. (R)-Limonene bioconversion to (R)-perillic acid has also been demonstrated in yeast Yarrowia

lipolytica (Ferrara et al., 2013), Mycobacterium sp. HXN-1500, Rhodococcus sp. ALK2-C7 (Duetz et al., 2004),

recombinant E. coli with cym genes from Pseudomonas putida F1 (Mars et al., 2001) and Pseudomonas

gladioli (Cadwallader et al., 1989), but all of them with lower productivities than P. putida GS1. Other

oxyfunctionalizations of (R)-limonene with several different strains were reviewed by Duetz and co-workers

(Duetz et al., 2003).

The fact that the strain used in this bioprocess is a wild-type strain also imposes some constraints in the

process, since wild-type enzymes are in most cases far from being applicable in industrially relevant

conditions. Product and substrate inhibition at high concentrations is one of the problems they often face.

Process technologies, such as ISPR or ISSS can partially solve these problems but only to a certain extent.

There is a limit on how much a process can be improved solely by process technologies and it might not be

enough to achieve commercial production. Most examples of biosynthetic processes for the production of

natural products that have achieved commercial scale include genetic improvement of the biocatalyst and/or

strain (Paddon and Keasling, 2014). But on the other hand, the use of a GMO strain might also have some

implications particularly on the marketing and consumer’s acceptance of the product.

121

5.3 Case-study II: Fermentation of recombinant Pichia pastoris for brazzein

production

In the second case-study only the first part of the proposed methodology, represented in Figure 5.2, was

applied. In case-study II, referring to the production of brazzein by fermentation of Pichia pastoris, only the

benchmarking of the biosynthetic route with other possible routes was addressed and the process was not

further evaluated or developed.

Figure 5.2 Part 1 of the proposed methodology for the development of biosynthetic processes for natural products production.

Besides benchmarking of different routes for brazzein production, the fermentation of recombinant Pichia

pastoris was also benchmarked against processes for production of different sweeteners, namely stevia and

aspartame. Even though these two processes are in different development stages than the fermentation

process and are already implemented in industrial scale, using them as benchmarks allowed the definition of

development targets for the teams working in process R&D. For instance, if brazzein aims to be

environmentally and economically competitive with stevia, PMI of the fermentation process needs to be

reduced to a value in the range of 100 kg/kg, which requires a product concentration of 35 g/L and a

downstream processing efficiency of 80%. This means that product concentration achieved in the

fermentation needs to be increase by approximately 17.5 fold and DSP efficiency needs to be improved from

122

approximately 30% to 80%. These might seem very ambitious goals, but setting this type of targets can guide

and focus research, which can ultimately result in shorter development time.

The results obtained from the assessment of this case-study illustrate the importance of addressing

environmental issues from an early time point on process research and development. Even though

economics is what drives process development, an environmental assessment can also be meaningful and

give valuable insight for process research and development. For instance, an assessment of the EHS issues of

the chemicals used can give suggestions for chemicals substitution that can greatly benefit the environmental

profile of the bioprocess. And this type of questions need to be addressed in an early stage of research and

development, otherwise it will be costly to make any change in the process.

Reporting environmental data, such as PMI, RME or E-factor, of processes at lab-scale research can also

benefit process development and research since it can provide some benchmarking examples for researchers

working on similar processes. This thesis reports environmental data for two bioprocesses for biosynthesis

of natural products, which is something that is not often seen in scientific literature.

The methodology followed and the metrics used to assess the environmental profile of this bioprocess

proved to be simple and easy to be applied by any researcher, even researchers who are not familiar with

green chemistry metrics or other tools for environmental assessment of bioprocesses in general. A full LCA

is not an appropriate method for environmental assessment of bioprocesses in early stages of research since

it is too complex and requires a lot of data that is difficult to obtain in early stages of research. The strength

of a LCA is the fact that it provides a holistic view of the environmental profile of a given product, taking into

account the raw materials production, transportation and the product use and disposal. However, from a

process development perspective that might not provide any meaningful information. Very often, a full LCA

“dilutes” the environmental profile of the specific process under development, since its emissions or waste

produced are insignificant when compared to energy production processes or transportation of the raw

materials. Therefore, in general terms LCA is not recommended for processes under development. It does

not support the process developer and it is rather complex and dependent on the experience of the user.

123

5.4 General recommendations for implementation of processes for the

biosynthesis of natural products

Estimates for the total number of living species ranges from 2 million to 100 million. Therefore, the number

of natural products and enzymes remaining to be discovered is enormous and there is a great potential for

discovery of molecules with very interesting properties. Synthetic biology shows the potential to map the

metabolism of most species and a library of biochemical transformations could be a magnificent tool for the

design and generation of new products. Just as organic chemists currently plan the total synthesis of a target

molecule using established reagents and reactions, synthetic biologists might be able in a near future to use

a vast array of enzymes to rationally synthesize very complex molecules. Directed evolution of enzymes or

ground-breaking gene editing technologies, such as CRISPR-Cas9, can greatly favor the biosynthesis of novel

natural products. The resources for natural products discovery are so vast that it is likely to see very exciting

discoveries in the field of biosynthetic production of natural products (Li and Vederas, 2009).

However, the development path until industrial scale and full commercialization is rather long. This thesis

proposes a methodology for addressing economic as well as environmental issues from an early stage of

research and development that can greatly benefit the implementation of more of these bioprocesses for

production of natural products. And knowledge sharing between different fields, such as synthetic biology

and chemical/biochemical engineering, is also crucial for a successful and fast development of these

biosynthetic processes.

124

125

6. Conclusions

Natural products show very interesting properties for use in several sectors of chemical industry, namely

pharmaceutical, food or cosmetics. However, they are extremely difficult to synthesize given their complex

structure. Synthetic biology offers tools with great potential for the synthesis of natural products and many

examples of biosynthesis of natural products, either by fermentation of recombinant strains or biocatalytic

reactions, have been reported in literature. But further development to industrial scale and

commercialization requires high yields, productivities and product concentrations. The intensification of

these bioprocesses is not straightforward and it is often a time-consuming process.

The main goal of this thesis was to establish a methodology for benchmarking of processes for the

biosynthesis of natural products that can assist the development of these processes. The work developed in

this thesis allowed drawing some general conclusions regarding the biosynthesis of natural products:

In contrast to other molecules, natural products can be produced by different means – extraction of

natural sources, chemical synthesis or biosynthesis. Route selection should be the first step of

development of a process for natural products production.

In case the biosynthetic route is the most promising option for production of a certain natural

product, an in silico evaluation of the economic and environmental profile of the process can help

setting targets for process development and identify the main bottlenecks of the process.

The proposed methodology demonstrated that that space-time yield was the biggest economic

constraint in the first case-study – the bioconversion of (R)-limonene to (R)-perillic acid by

Pseudomonas putida GS1. A 4-fold improvement of space-time yield could result in a reduction of

40% on operation costs, which is a very significant reduction. The increase of space-time yield would

also result in an improved environmental profile of the process given the interconnection of

economic and environmental performance.

Substrate loss by stripping on the gas stream and a reduction of reaction rate throughout reaction

time due to product inhibition were identified as some of the causes for a low space-time yield.

In the second case-study, the fermentation of recombinant Pichia pastoris for brazzein production

was benchmarked with the production processes of two other sweeteners, stevia and aspartame,

from an environmental perspective. The environmental assessment proved to be a valuable tool to

define targets for the development of the fermentation process. A sensitivity analysis on some

process parameters showed that if product concentration is increased to approximately 35 g/L and

126

DSP efficiency to 80% a PMI of 115 kg/kg could be achieved, which would make the fermentation

process for brazzein production competitive with stevia production.

The in silico benchmarking and evaluation of biosynthetic processes for production of natural

products is a valuable tool to define a selling price for a novel product without an identical competitor

in the market.

127

7. Future work

To further explore the process development of biosynthetic processes for natural products production, some

suggestions that would improve this thesis are presented. Guidelines for the development of both case-

studies are followed by some general guidelines for the development of processes for the biosynthesis of

natural products.

Bioconversion of (R)-limonene to (R)-perillic acid by a whole-cell catalyst

The reaction system should be further explored experimentally in order to understand the reason(s)

for the reduction in reaction rate that was observed. There are a lot of factors possibly influencing

the reaction rate (cofactor regeneration problems, loss of enzyme activity, transport of substrate and

product across the cell membrane, etc.) and they should be tested experimentally.

The operating mode of the biocatalytic reaction should be further explored.

The ISPR operating mode should be further developed.

The application of a second organic phase that can be used as a substrate reservoir should be tested.

This could diminish the substrate loss through stripping on the gas phase and potentially also enable

ISPR. Examples where this was successfully applied are available in literature for similar systems

(Schewe et al., 2009). Cornelissen and co-workers showed an improvement in productivity when

running the bioconversion of (S)-limonene to (S)-perillyl alcohol by a recombinant Pseudomonas

putida strain in a two-liquid-phase system (Cornelissen et al., 2011).

Since the biocatalyst is a whole-cell there might be some transport limitations of the substrate and/or

product across the cell membrane. The reaction rate inside the cell might actually be higher than

what was measured experimentally. If there are transport limitations in the membrane a lot of

product might be held inside the cell. This should be tested by disrupting the cells after the

bioconversion and measure (R)-perillic acid concentration in the cell lysate.

The process model could be improved with real price quotes from suppliers. Since facility-dependent

costs were one of the biggest contributors to the costing of the first case-study, it would be

interesting to get a real quote from a manufacturing facility for renting their equipment, thereby

improving the accuracy of the economic assessment and sensitivity analysis.

128

Fermentation of recombinant Pichia pastoris for brazzein production

In this case-study both upstream and downstream processing need to be improved. Since it is a

recombinant strain that is used further metabolic engineering of the strain needs to be done in order

to increase the product concentration in the fermentation.

Optimization of the operating conditions in DSP should be done in order to improve the efficiency of

DSP. Design of experiments is a tool that can help and accelerate this experimental work.

General guidelines for the development of processes for the biosynthesis of natural products

The development of more in silico tools for predicting properties at operating conditions could

reduce experimental work and facilitate and accelerate process development. Natural products are

complex molecules, whose properties, e.g. solubility, pKa or log P, are not easily found in scientific

literature. Experimentally measuring all these physicochemical properties is a time-consuming

process. Therefore, having predictive tools for prediction of pKa, partition coefficients, etc. could

reduce experimental work and assist process design and development.

The application of design of experiments tools can also be a valuable way to acquire more process

knowledge and could benefit the development of these biosynthetic processes.

129

8. References

Aharoni, A., Jongsma, M.A., Bouwmeester, H.J., 2005. Volatile science? Metabolic engineering of terpenoids

in plants. Trends Plant Sci. 10, 594–602.

Ajikumar, P.K., Xiao, W.-H., Tyo, K.E.J., Wang, Y., Simeon, F., Leonard, E., Mucha, O., Phon, T.H., Pfeifer, B.,

Stephanopoulos, G., 2010. Isoprenoid pathway optimization for Taxol precursor overproduction in

Escherichia coli. Science 330, 70–4.

Alfonsi, K., Colberg, J., Dunn, P.J., Fevig, T., Jennings, S., Johnson, T.A., Kleine, H.P., Knight, C., Nagy, M.A.,

Perry, D.A., Stefaniak, M., 2008. Green chemistry tools to influence a medicinal chemistry and research

chemistry based organisation. Green Chem. 10, 31–36.

Anastas, P., Eghbali, N., 2010. Green chemistry: principles and practice. Chem. Soc. Rev. 39, 301–312.

Anthony, J.R., Anthony, L.C., Nowroozi, F., Kwon, G., Newman, J.D., Keasling, J.D., 2009. Optimization of the

mevalonate-based isoprenoid biosynthetic pathway in Escherichia coli for production of the anti-

malarial drug precursor amorpha-4,11-diene. Metab. Eng. 11, 13–19.

Arnold, F.H., 2001. Combinatorial and computational challenges for biocatalyst design. Nature 409, 253–257.

Assadi-Porter, F.M., Aceti, D.J., Cheng, H., Markley, J.L., 2000. Efficient production of recombinant brazzein,

a small, heat-stable, sweet-tasting protein of plant origin. Arch. Biochem. Biophys. 376, 252–8.

Berger, R.G. (Ed.), 2007. Flavours and Fragrances: Chemistry, Bioprocessing and Sustainability, Journal of

Chemical Information and Modeling. Springer.

Bornscheuer, U.T., Huisman, G.W., Kazlauskas, R.J., Lutz, S., Moore, J.C., Robins, K., 2012. Engineering the

third wave of biocatalysis. Nature 485, 185–194.

Brunet, R., Guillén-Gosálbez, G., Jiménez, L., 2012. Cleaner design of single-product biotechnological facilities

through the integration of process simulation, multiobjective optimization, life cycle assessment, and

principal component analysis. Ind. Eng. Chem. Res. 51, 410–424.

Byrne, F.P., Jin, S., Paggiola, G., Petchey, T.H.M., Clark, J.H., Farmer, T.J., Hunt, A.J., Robert McElroy, C.,

Sherwood, J., 2016. Tools and techniques for solvent selection: green solvent selection guides. Sustain.

Chem. Process. 4, 7.

Cadwallader, K.R., Braddock, R.J., Parish, M.E., Higgins, D.P., 1989. Bioconversion of (+)-Limonene by

130

Pseudomonas gladioli. J. Food Sci. 54, 1241–1245.

Chotani, G., Dodge, T., Hsu, A., Kumar, M., LaDuca, R., Trimbur, D., Weyler, W., Sanford, K., 2000. The

commercial production of chemicals using pathway engineering. Biochim. Biophys. Acta - Protein Struct.

Mol. Enzymol. 1543, 434–455.

Constable, D.J.C., Curzons, A.D., Cunningham, V.L., 2002. Metrics to “green” chemistry - which are the best?

Green Chem. 4, 521–527.

Constable, D.J.C., Jimenez-Gonzalez, C., Henderson, R.K., 2007. Perspective on solvent use in the

pharmaceutical industry. Org. Process Res. Dev. 11, 133–137.

Cornelissen, S., Liu, S., Deshmukh, A.T., Schmid, A., Bühler, B., 2011. Cell physiology rather than enzyme

kinetics can determine the efficiency of cytochrome P450-catalyzed C-H-oxyfunctionalization. J. Ind.

Microbiol. Biotechnol. 38, 1359–1370.

Curzons, A.D., Jiménez-González, C., Duncan, A.L., Constable, D.J.C., Cunningham, V.L., 2007. Fast life cycle

assessment of synthetic chemistry (FLASCTM) tool. Int. J. Life Cycle Assess. 12, 272–280.

Curzons, A.D., Mortimer, D.N., Constable, D.J.C., Cunningham, V.L., 2001. So you think your process is green,

how do you know? — Using principles of sustainability to determine what is green – a corporate

perspective. Green Chem. 3, 1–6.

Das, A., Yoon, S.H., Lee, S.H., Kim, J.Y., Oh, D.K., Kim, S.W., 2007. An update on microbial carotenoid

production: Application of recent metabolic engineering tools. Appl. Microbiol. Biotechnol. 77, 505–

512.

de María, P.D., Hollmann, F., 2015. On the (Un)greenness of biocatalysis: Some challenging figures and some

promising options. Front. Microbiol. 6, 6–10.

Duetz, W.A., Bouwmeester, H., Van Beilen, J.B., Witholt, · B, 2003. Biotransformation of limonene by bacteria,

fungi, yeasts, and plants. Appl Microbiol Biotechnol 61, 269–277.

Duetz, W.A., Witholt, B., Jourdat, C., 2004. Process for the preparation of perillyl alcohol. Patent No.

US2004/0077063 A1.

Dunn, P.J., 2012. The importance of Green Chemistry in Process Research and Development. Chem. Soc. Rev.

41, 1452–1461.

Dunn, P.J., Hettenbach, K., Kelleher, P., Martinez, C.A., 2010. The Development of a Green, Energy Efficient,

131

Chemoenzymatic Manufacturing Process for Pregabalin. In: Dunn, P.J., Wells, A.S., Williams, M.T. (Eds.),

Green Chemistry in the Pharmaceutical Industry. Wiley-VCH Verlag GmbH & Co. KGaA, pp. 161–177.

Engels, B., Dahm, P., Jennewein, S., 2008. Metabolic engineering of taxadiene biosynthesis in yeast as a first

step towards Taxol (Paclitaxel) production. Metab. Eng. 10, 201–206.

European Comission for Standardization, 2006a. ISO 14040:2006 - Environmental management - Life cycle

assessment - Principles and framework.

European Comission for Standardization, 2006b. ISO 14044:2006 Environmental Management – Life Cycle

Assessment – Requirements and Guidelines.

European Commission, 2013. A Bioeconomy Strategy for Europe.

European Commission, 2014. Where next for the European bioeconomy?

European Food Safety Authority (EFSA), 2010. Scientific Opinion on the safety of steviol glycosides for the

proposed uses as a food additive. EFSA J. 8, 1–84.

Fatima, A., Terra, B., Silva, C., Silva, D., Araujo, D., Neto, L., Aquino, R., 2014. From Nature to Market: Examples

of Natural Products that Became Drugs. Recent Pat. Biotechnol. 8, 76–88.

Federal Statistical Office of Germany, 2016. Feldfrüchte und Grünland [WWW Document]. URL

https://www.destatis.de/DE/ZahlenFakten/Wirtschaftsbereiche/LandForstwirtschaftFischerei/Feldfru

echteGruenland/Tabellen/FeldfruechteZeitreihe.html (accessed 8.1.15).

Ferrara, M.A., Almeida, D.S., Siani, A.C., Lucchetti, L., Lacerda, P.S.B., Freitas, A., Tappin, M.R.R., Bon, E.P.S.,

2013. Bioconversion of R- ( + ) -limonene to perillic acid by the yeast Yarrowia lipolytica. Brazilian J.

Microbiol. 1080, 1075–1080.

Frense, D., 2007. Taxanes: Perspectives for biotechnological production. Appl. Microbiol. Biotechnol. 73,

1233–1240.

Fujita, T., Nakayama, M., 1993. Monoterpene glucosides and other constituents from Perilla frutescens.

Phytochemistry 34, 1545–1548.

Gerbens-Leenes, P.W., Hoekstra, A.Y., 2009. The water footprint of sweeteners and bio-ethanol from sugar

cane, sugar beet and maize, UNESCO-IHE, Institute for Water Education.

Golden, J.S., Handfield, R., 2014. The Emergent Industrial Bioeconomy. Ind. Biotechnol. 10, 371–375.

Hanzawa, S., 2010. Aspartame. Encycl. Ind. Biotechnol. Bioprocess, Biosep. Cell Technol. 1–9.

132

Harrison, R.G., Todd, P.W., Rudge, S.R., Petrides, D.P., 2015. Bioseparations Science and Engineering. Oxford

University Press.

Hawkins, K.M., Smolke, C.D., 2008. Production of benzylisoquinoline alkaloids in Saccharomyces cerevisiae.

Nat. Chem. Biol. 4, 564–573.

Heinzle, E., Biwer, A.P., Cooney, C.L., 2008. Development of Sustainable Bioprocesses: Modelling and

Assessment. John Wiley & Sons, Ltd.

Heinzle, E., Hungerbuhler, K., 1997. Integrated Process Development: The Key to Future Production of

Chemicals. Chim. Int. J. Chem. 51, 176–183.

Heinzle, E., Weirich, D., Brogli, F., Hoffmann, V.H., Koller, G., Verduyn, M. a., Hungerbühler, K., 1998.

Ecological and Economic Objective Functions for Screening in Integrated Development of Fine Chemical

Processes. 1. Flexible and Expandable Framework Using Indices. Ind. Eng. Chem. Res. 37, 3395–3407.

Heinzle E.; Weirich, D.. B.F. ;Hoffman. V.H.. K.G.V.M.A., K., H., 1998. Ecological and Economic Objective

Funcions for Screenig in Integrated Development of Fine Chemical Processes. 2. Stream Allocation and

Case Studies. Ind. Eng. Chem. Res 37, 3408–3413.

Hellekant, B.G., Ming, D., 1998. Brazzein sweetener. Patent No. 5,741,537.

Henderson, R.K., Jiménez-González, C., Constable, D.J.C., Alston, S.R., Inglis, G.G. a., Fisher, G., Sherwood, J.,

Binks, S.P., Curzons, A.D., 2011. Expanding GSK’s solvent selection guide – embedding sustainability into

solvent selection starting at medicinal chemistry. Green Chem. 13, 854.

Hudlicky, T., Frey, D.A., Koroniak, L., Claeboe, C.D., Larry, E., 1999. Toward a “ reagent-free ” synthesis. Green

Chem. 57–59.

Jimenez-Gonzalez, C., Ponder, C.S., Broxterman, Q.B., Manley, J.B., 2011. Using the right green yardstick: Why

process mass intensity is used in the pharmaceutical industry to drive more sustainable processes. Org.

Process Res. Dev. 15, 912–917.

Katsuyama, Y., Funa, N., Miyahisa, I., Horinouchi, S., 2007. Synthesis of Unnatural Flavonoids and Stilbenes

by Exploiting the Plant Biosynthetic Pathway in Escherichia coli. Chem. Biol. 14, 613–621.

Kiyotaka Oyama, H., Shigeaki Irino, T., Tsuneo Harada, S., Masao Nakamura, Y., 1985. Process for producing

an addition compound of a dipeptide ester and an amino acid ester. Patent No. 4,521,514.

Kuhn, D., Julsing, M.K., Heinzle, E., Buhler, B., 2012. Systematic optimization of a biocatalytic two-liquid phase

133

oxyfunctionalization process guided by ecological and economic assessment. Green Chem. 14, 645–653.

Kuhn, D., Kholiq, M.A., Heinzle, E., Bühler, B., Schmid, A., 2010. Intensification and economic and ecological

assessment of a biocatalytic oxyfunctionalization process. Green Chem. 12, 815.

Lamphear, B.J., Barker, D.K., Brooks, C.A., Delaney, D.E., Lane, J.R., Beifuss, K., Love, R., Thompson, K., Mayor,

J., Clough, R., Harkey, R., Poage, M., Drees, C., Horn, M.E., Streatfield, S.J., Nikolovt, Z., Woodard, S.L.,

Hood, E.E., Jilka, J.M., Howard, J.A., 2005. Expression of the sweet protein brazzein in maize for

production of a new commercial sweetener. Plant Biotechnol. J. 3, 103–114.

Lapkin, A., Constable, D.J.C., 2009. Green Chemistry Metrics: Measuring and Monitoring Sustainable

Processes, Green Chemistry Metrics: Measuring and Monitoring Sustainable Processes. John Wiley &

Sons, Ltd.

Leahy, D.K., Tucker, J.L., Mergelsberg, I., Dunn, P.J., Kopach, M.E., Purohit, V.C., 2013. Seven important

elements for an effective green chemistry program: An IQ consortium perspective. Org. Process Res.

Dev. 17, 1099–1109.

Lee, P.C., Schmidt-Dannert, C., 2003. Metabolic engineering towards biotechnological production of

carotenoids in microorganisms. Appl. Microbiol. Biotechnol. 60, 1–11.

Li, J.W.-H., Vederas, J.C., 2009. Drug discovery and natural products: end of an era or an endless frontier?

Science 325, 161–5.

Liang, J., Mundorff, E., Voladri, R., Jenne, S., Gilson, L., Conway, A., Krebber, A., Wong, J., Huisman, G.,

Truesdell, S., Lalonde, J., 2010. Highly Enantioselective Reduction of a Small Heterocyclic Ketone:

Biocatalytic Reduction of Tetrahydrothiophene-3-one to the Corresponding (R)-Alcohol. Org. Process

Res. Dev. 14, 188–192.

Lima-Ramos, J., Neto, W., Woodley, J.M., 2014a. Engineering of Biocatalysts and Biocatalytic Processes. Top.

Catal. 57, 301–320.

Lima-Ramos, J., Tufvesson, P., Woodley, J.M., 2014b. Application of environmental and economic metrics to

guide the development of biocatalytic processes. Green Process. Synth. 3, 195–213.

Lindeberg, G., 1987. A convenient synthesis of aspartame. J. Chem. Educ. 64, 1062.

Lundemo, M.T., Woodley, J.M., 2015. Guidelines for development and implementation of biocatalytic P450

processes. Appl. Microbiol. Biotechnol. 99, 2465–2483.

134

Ma, S., Wang, A., 2012. Molecular farming in plants: An overview, Molecular Farming in Plants: Recent

Advances and Future Prospects. Springer.

Ma, S.K., Gruber, J., Davis, C., Newman, L., Gray, D., Wang, A., Grate, J., Huisman, G.W., Sheldon, R. a., 2010.

A green-by-design biocatalytic process for atorvastatin intermediate. Green Chem. 12, 81.

Mars, A.E., Gorissen, J.P.L., Van den Beld, I., Eggink, G., 2001. Bioconversion of limonene to increased

concentrations of perillic acid by Pseudomonas putida GS1 in a fed-batch reactor. Appl. Microbiol.

Biotechnol. 56, 101–107.

Martin, V.J.J., Pitera, D.J., Withers, S.T., Newman, J.D., Keasling, J.D., 2003. Engineering a mevalonate

pathway in Escherichia coli for production of terpenoids. Nat. Biotechnol. 21, 796–802.

McElroy, C.R., Constantinou, A., Jones, L.C., Summerton, L., Clark, J.H., 2015. Towards a holistic approach to

metrics for the 21st century pharmaceutical industry. Green Chem. 17, 3111–3121.

Menkhaus, T.J., Bai, Y., Zhang, C., Nikolov, Z.L., Glatz, C.E., 2004. Considerations for the recovery of

recombinant proteins from plants. Biotechnol. Prog. 20, 1001–1014.

Midmore, D., Rank, A., 2002. A new rural industry – Stevia – to replace imported chemical sweeteners. Rural

Ind. Res. Dev. Corp. 1–55.

Mikkelsen, M.D., Hansen, J., Simon, E., Brianza, F., Semmler, A., Olsson, K., Carlsen, S., Düring, L., Ouspenski,

A., Hicks, P., 2014. Methods for improved production of rebaudioside D and rebaudioside M. Patent No.

WO/2014/122227.

Minami, H., Kim, J.-S., Ikezawa, N., Takemura, T., Katayama, T., Kumagai, H., Sato, F., 2008. Microbial

production of plant benzylisoquinoline alkaloids. Proc. Natl. Acad. Sci. U. S. A. 105, 7393–7398.

Mirata, M.A., Heerd, D., Schrader, J., 2009. Integrated bioprocess for the oxidation of limonene to perillic

acid with Pseudomonas putida DSM 12264. Process Biochem. 44, 764–771.

Mitchell, H., 2006. Sweeteners and sugar alternatives in food technology. Blackwell Publishing Ltd.

Newman, D.J., Cragg, G.M., 2007. Natural products as sources of new drugs over the last 25 years. J. Nat.

Prod. 70, 461–477.

Newman, D.J., Cragg, G.M., Snader, K.M., 2003. Natural products as sources of new drugs over the period

1981-2002. J. Nat. Prod. 66, 1022–1037.

Ni, Y., Holtmann, D., Hollmann, F., 2014. How Green is Biocatalysis? To Calculate is To Know. ChemCatChem

135

6, 930–943.

Nicolaou, K.C., Yang, Z., Liu, J.J., Ueno, H., Nantermet, P.G., Guy, R.K., Claiborne, C.F., Renaud, J., Couladouros,

E.A., Paulvannan, K., 1994. Total synthesis of taxol. Nature 367, 630–4.

Oudshoorn, a., van den Berg, C., Roelands, C.P.M., Straathof, a. J.J., van der Wielen, L. a. M., 2010. Short-

cut calculations for integrated product recovery options in fermentative production of bio-bulk

chemicals. Process Biochem. 45, 1605–1615.

Paddon, C.J., Keasling, J.D., 2014. Semi-synthetic artemisinin: a model for the use of synthetic biology in

pharmaceutical development. Nat. Rev. Microbiol. 12, 355–367.

Pearsall, S.M., Rowley, C.N., Berry, A., 2015. Advances in Pathway Engineering for Natural Product

Biosynthesis. ChemCatChem 7, 3078–3093.

Perry, R., Green, D., Maloney, J., 1997. Perry’s chemical engineers’ handbook. McGraw Hill.

Peters, M.S., Timmerhaus, K.D., West, R.E., 2004. Plant Design and Economics for Chemical Engineers.

McGraw Hill.

Pfeifer, B.A., Admiraal, S.J., Gramajo, H., Cane, D.E., Khosla, C., 2001. Biosynthesis of Complex Polyketides in

a Metabolically Engineered Stain of E. coli. Science 291, 1790–1792.

Prat, D., Hayler, J., Wells, A., 2014. A Survey of Solvent Selection Guides. Green Chem. 16, 4546–4551.

Prat, D., Pardigon, O., Flemming, H.-W., Letestu, S., Ducandas, V., Isnard, P., Guntrum, E., Senac, T., Ruisseau,

S., Cruciani, P., Hosek, P., 2013. Sanofi’s Solvent Selection Guide: A Step Toward More Sustainable

Processes. Org. Process Res. Dev. 1517–1525.

Prat, D., Wells, A., Hayler, J., Sneddon, H., McElroy, C.R., Abou-Shehada, S., Dunn, P.J., 2016. CHEM21

selection guide of classical- and less classical-solvents. Green Chem. 18, 288–296.

Puri, M., Sharma, D., Tiwari, A.K., 2011. Downstream processing of stevioside and its potential applications.

Biotechnol. Adv. 29, 781–791.

Purkayastha, S., Markosyan, A., Malsagov, M., 2012. Process for manufacturing a sweetener and use thereof.

Patent No. 8,337,925 B2.

Ramesh, H., Nordblad, M., Whittall, J., Woodley, J.M., 2016. Considerations for the Application of Process

Technologies in Laboratory- and Pilot-Scale Biocatalysis for Chemical Synthesis. In: Whittall, J., Sutton,

P.W., Kroutil, W. (Eds.), Practical Methods for Biocatalysis and Biotransformations 3. John Wiley & Sons,

136

Ltd.

Rieks, A., Kähler, M., Kirchner, U., Wiggenhorn, K., Kinzer, M., Risch, S., 2004. Antimicrobial agents against

bacteria, yeast and moulds. Patent No. WO2004076400.

Roschangar, F., Sheldon, R.A., Senanayake, C.H., 2015. Overcoming barriers to green chemistry in the

pharmaceutical industry – the Green Aspiration LevelTM concept. Green Chem. 17, 752–768.

Savile, C.K., Janey, J.M., Mundorff, E.C., Moore, J.C., Tam, S., Jarvis, W.R., Colbeck, J.C., Krebber, A., Fleitz,

F.J., Brands, J., Devine, P.N., Huisman, G.W., Hughes, G.J., 2010. Biocatalytic Asymmetric Synthesis of

Sitagliptin Manufacture. Science 329, 305–310.

Schewe, H., Holtmann, D., Schrader, J., 2009. P450BM-3-catalyzed whole-cell biotransformation of α-pinene

with recombinant Escherichia coli in an aqueous-organic two-phase system. Appl. Microbiol.

Biotechnol. 83, 849–857.

Schmid, A., Dordick, J.S., Hauer, B., Kiener, A., Wubbolts, M., Witholt, B., 2001. Industrial biocatalysis today

and tomorrow. Nature 409, 258–268.

Sheldon, R.A., 1992. Organic synthesis; past, present and future. Chem. Ind. 903–906.

Sheldon, R.A., 2011. Utilisation of biomass for sustainable fuels and chemicals: Molecules, methods and

metrics. Catal. Today 167, 3–13.

Shonnard, D.R., Kicherer, A., Saling, P., 2003. Industrial Applications Using BASF Eco-Efficiency Analysis:

Perspectives on Green Engineering Principles. Environ. Sci. Technol. 37, 5340–5348.

Speelmans, G., Bijlsma, A., Eggink, G., 1998. Limonene bioconversion to high concentrations of a single and

stable product, perillic acid, by a solvent-resistant Pseudomonas putida strain. Appl. Microbiol.

Biotechnol. 50, 538–544.

Straathof, A.J.J., 2014. Transformation of biomass into commodity chemicals using enzymes or cells. Chem.

Rev. 114, 1871–1908.

Straathof, A.J.J., Panke, S., Schmid, A., 2002. The production of fine chemicals by biotransformations. Curr.

Opin. Biotechnol. 13, 548–556.

Straathof, A.J.J., Adlercreutz, P., 2000. Applied Biocatalysis. CRC Press.

Sun, Z., Ramsay, J.A., Guay, M., Ramsay, B.A., 2006. Automated feeding strategies for high-cell-density fed-

batch cultivation of Pseudomonas putida KT2440. Appl Microbiol Biotechnol 71, 423–431.

137

Tao, J. (Alex), Kazlauskas, R. (Eds.), 2011. Biocatalysis for Green Chemistry and Chemical Process

Development. Wiley.

Thomas, S.M., DiCosimo, R., Nagarajan, V., 2002. Biocatalysis: Applications and potentials for the chemical

industry. Trends Biotechnol. 20, 238–242.

Tobiszewski, M., Tsakovski, S., Simeonov, V., Namieśnik, J., Pena-Pereira, F., 2015. A solvent selection guide

based on chemometrics and multicriteria decision analysis. Green Chem. 17, 4773–4785.

Trost, B.M., 1991. The Atom Economy-A Search for Synthetic Efficiency. Science 254, 1471–1477.

Truppo, M.D., Hughes, G., 2011. Development of an improved immobilized CAL-B for the enzymatic

resolution of a key intermediate to odanacatib. Org. Process Res. Dev. 15, 1033–1035.

Tsao, G.T., Lee, D.D., 1975. Oxygen transfer in fermentation. AIChE J. 21, 979–985.

Tufvesson, L.M., Tufvesson, P., Woodley, J.M., Börjesson, P., 2012. Life cycle assessment in green chemistry:

overview of key parameters and methodological concerns. Int. J. Life Cycle Assess. 431–444.

Tufvesson, P., Fu, W., Jensen, J.S., Woodley, J.M., 2010. Process considerations for the scale-up and

implementation of biocatalysis. Food Bioprod. Process. 88, 3–11.

Turner, N.J., 2009. Directed evolution drives the next generation of biocatalysts. Nat. Chem. Biol. 5, 567–573.

US Environmental Protection Agency, n.d. Presidential Green Chemistry Challenge Winners [WWW

Document]. URL https://www.epa.gov/greenchemistry/presidential-green-chemistry-challenge-

winners (accessed 7.29.16).

van’t Riet, K., 1986. The selection of research targets based on forecasted production costs. Trends

Biotechnol. 4, 236–241.

Van Dien, S., 2013. From the first drop to the first truckload: Commercialization of microbial processes for

renewable chemicals. Curr. Opin. Biotechnol. 24, 1061–1068.

Vellinga, T., Boer, J. de, 2012. LCI data for the calculation tool Feedprint for greenhouse gas emissions of feed

production and utilization - Cultivation of forage and roughage. Blonk Consultants.

Villadsen, J., 2015. Fundamental Bioengineering. Wiley.

von Kitzing, E., Jonas, P., Sakmann, B., 1994. Effect of naturally occurring monoterpenes carvone, limonene

and perillic acid in the inhibition of experimental lung metastasis induced by B16F-10 melanoma cells.

Res. Vet. Sci. 30, 364–367.

138

Walters, D.E., Orthoefer, F.T., DuBois, G.E., of Agricultural, A.C.S.D., Chemistry, F., Meeting, A.C.S., 1991.

Sweeteners: discovery, molecular design, and chemoreception, ACS symposium series. American

Chemical Society.

Wang, Q., Fan, S.Y., Wang, H.N.C., 1993. Enantioselective Synthesis of Chiral Liquid Crystalline Compounds

from Monoterpenes. Tetrahedron 49, 619–638.

Weiss, M., Haufe, J., Carus, M., Brandão, M., Bringezu, S., Hermann, B., Patel, M.K., 2012. A Review of the

Environmental Impacts of Biobased Materials. J. Ind. Ecol. 16.

Wenda, S., Illner, S., Mell, A., Kragl, U., 2011. Industrial biotechnology—the future of green chemistry? Green

Chem. 13, 3007.

Woodley, J.M., 2008. New opportunities for biocatalysis: making pharmaceutical processes greener. Trends

Biotechnol. 26, 321–327.

Woodley, J.M., Breuer, M., Mink, D., 2013. A future perspective on the role of industrial biotechnology for

chemicals production. Chem. Eng. Res. Des. 91, 2029–2036.

Wu, S., Chappell, J., 2008. Metabolic engineering of natural products in plants; tools of the trade and

challenges for the future. Curr. Opin. Biotechnol. 19, 145–152.

Xie, X., Watanabe, K., Wojcicki, W.A., Wang, C.C.C., Tang, Y., 2006. Biosynthesis of Lovastatin Analogs with a

Broadly Specific Acyltransferase. Chem. Biol. 13, 1161–1169.

Yadav, V.G., De Mey, M., Giaw Lim, C., Kumaran Ajikumar, P., Stephanopoulos, G., 2012. The future of

metabolic engineering and synthetic biology: Towards a systematic practice. Metab. Eng. 14, 233–241.

Yeruva, L., Pierre, K.J., Elegbede, A., Wang, R.C., Carper, S.W., 2007. Perillyl alcohol and perillic acid induced

cell cycle arrest and apoptosis in non small cell lung cancer cells. Cancer Lett. 257, 216–226.

Yu, H., Kosuna, K., Haga, M. (Eds.), 1997. Perilla - The Genus Perilla. Taylor&Francis Group.

Zorn, H., Czermark, P., 2014. Biotechnology of Food and Feed Additives, Advances in biochemical

engineering/biotechnology. Springer-Verlag Berlin Heidelberg.

i

Appendices

Mass balances for the chemical route for (R)-perillic acid synthesis from β-pinene

- The data on reagents consumption for the chemical synthesis of perillic acid was based on (Wang et

al., 1993).

Reaction Step 1

Reaction yield: 35% (based on benzonitrile)

Input Mass (g) Output Mass (g)

β-pinene 5.88 β-pinene epoxide 2.24

H2O2 (30%) 9.52 Unreacted reagents 17.44

Benzonitrile (PhCN) 4.28

Sum 19.68 Sum 19.68

Reaction Step 2

Reaction yield: 65%

Input Mass (g) Output Mass (g)

β-pinene epoxide 2.24 Perillyl alcohol 1.52

NH4NO3 0.15 Unreacted reagents 0.87

Sum 2.39 Sum 2.39

ii

Reaction Steps 3 and 4

Overall reaction yield of 77% from perillyl alcohol

Input Mass (g) Output Mass (g)

Perillyl alcohol 1.52 Perillic acid 1.28

MnO2 17.4 Unreacted reagents 17.64

AgNO3 1.7

NaOH (aq.) 0.40

Sum 18.92 Sum 18.92

iii

Calibration curve for HPLC measurements of (R)-perillic acid

y = 4,5801x - 0,0725R² = 0,9988

0

1

2

3

4

5

6

7

8

9

10

0 0,5 1 1,5 2

Pe

ak a

rea

(mA

U*m

in)

[(R)-perillic acid] (mM)

iv

Calibration curve for the UV-VIS measurements of (R)-limonene at 254 nm

y = 3,1931x - 0,0463R² = 0,9996

0

1

2

3

4

5

6

0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 1,8

Ab

sorp

tio

n

Limonene concentration (% (v/v))

v

Mass balances for the processes assessed in case-study II

1. Brazzein extraction from the fruits of Pentadiplandra brazzeana Baillon

Separation of the pulp from the epicarp and seeds and solid-liquid extraction

Input Mass (g) Output Mass (g)

Fruits of Pentadiplandra

brazzeana 2,000

Solution after the

extraction 43,364

Sodium acetate 164 Fruit pulp (Waste) 1,000

Dithiothreitol 0.6

Glycerol 2,000

Polyvinylpyrrolidone (PVP) 200

Water 40,000

Sum 44,364.6 Sum 44,364

Purification of the extraction solution

Input Mass (g) Output Mass (g)

Extraction solution 43,364 Product 0.9

Ammonium sulfate 24,682 Waste 70,914.6

Sodium hydroxide 1.4

Sodium chloride 51.6

Sodium acetate 3

Tris hydrochloride 0.5

Water 1,813

Sum 70,915.5 Sum 70,915.5

vi

2. Brazzein fermentation

Fermentation

Input Mass (kg) Output Mass (kg)

Glycerol 770.7

Methanol 2,313

Monopotassium

phosphate 236 Fermentation broth 10,304.6

Ammonium sulfate 27.5

Calcium chloride 5.5

Potassium sulfate 78.7

Magnesium sulfate 64.4

Copper sulfate 1.1x10-2

Sodium iodide 4.4x10-4

Manganese sulfate 1.65x10-2

Sodium molybdate 1.1x10-3

Boric acid 1.1x10-4

Calcium sulfate 2.75x10-3

Cobalt chloride 2.75x10-3

Zinc chloride 3.85x10-2

Iron sulfate 0.121

Biotin 1.1x10-3

Sulfuric acid 9.9x10-6

Inoculum 500

Phosphoric acid 8.6

Ammonia 154

Antifoam (Propylene

glycol) 5

Water 6,141.2

Sum 10,304.6 Sum 10,304.6

vii

Cell removal

Input Mass (kg) Output Mass (kg)

Fermentation broth 10,304.6 Cells 4,120.5

Supernatant 6,184.1

Sum 10,304.6 Sum 10,304.6

Clarification

Input Mass (kg) Output Mass (kg)

Supernatant 6,184.1 Waste 156.8

Water 250 Clarification solution 6,275.3

Sum 6,434.1 Sum 6,434.1

Concentration

Input Mass (kg) Output Mass (kg)

Clarification solution 6,275.3 Waste 4,215.7

Concentration solution 2,059.6

Sum 6,275.3 Sum 6,275.3

Buffer exchange

Input Mass (kg) Output Mass (kg)

Concentration solution 2,059.6 Waste 18,094.9

Sodium acetate 14.76 Solution 2,032

Sodium chloride 52.56

Water 18,000

Sum 20,126.9 Sum 20,126.9

viii

Chromatography

Input Mass (kg) Output Mass (kg)

Solution 2,032 Waste 6,062.4

Sodium acetate 4.2 Elution solution 1,300

Sodium chloride 126.1

Water 5,200

Sum 7,362.4 Sum 7,362.4

Freeze Drying

Input Mass (kg) Output Mass (kg)

Elution solution 1,300 Product 6

Waste 1,294

Sum 1,300 Sum 1,300

ix

3. Brazzein extraction from GMO maize

Cultivation of 1 hectare of maize

Input Mass (kg) Output Mass (kg)

Nitrogen fertilizer 30 Maize 3,500

Lime 400

Manure 170

Pesticides 0.9

Sum 600.9 Sum 3,500

Dry milling

Input Mass (kg) Output Mass (kg)

Maize 3,500 Maize middlings (waste) 2,800

Germ-rich fraction 700

Sum 3,500 Sum 3,500

Solid-liquid extraction

Input Mass (kg) Output Mass (kg)

Maize, gem-rich fraction 700 Extraction solution 4,211.8

Sodium acetate 5.7

Sodium chloride 6.1

Water 3,500

Sum 4,211.8 Sum 4,211.8

x

Purification

Input Mass (kg) Output Mass (kg)

Extraction solution 4,211.8 Product 0.403

Sodium chloride 205 Waste 11,440.4

Water 7,024

Sum 11,440.8 Sum 11,440.8

xi

4. Stevia production

Cultivation of stevia in one hectare of land in one year

Input Mass (kg) Output Mass (kg)

Nitrogen fertilizer 40 Dried stevia leaves 1800

Phosphorus fertilizer 20

Potassium fertilizer 30

Sum 90 Sum 1800

Solid-liquid extraction

Input Mass (kg) Output Mass (kg)

Dried leaves 1,800 Extraction solution 19,800

Water 18,000

Sum 19,800 Sum 19,800

Preliminary purification

Input Mass (kg) Output Mass (kg)

Extraction solution 19,800 Product powder 1 220

Calcium hydroxide 43.2 Waste 19,719

Iron hydroxide 95.4

Sum 19,939 Sum 19,939

Preparation of Stevioside

Input Mass (kg) Output Mass (kg)

Product powder 1 220 Product powder 3 129

Methanol 867 Waste 1227

Ethanol 242

Water 26.9

Sum 1356 Sum 1356

xii

Preparation of Rebaudioside A

Input Mass (kg) Output Mass (kg)

Water 909.4 Product powder 6 52

Ethanol 255.3 Waste 1113

Sum 1165 Sum 1165

xiii

5. Aspartame production

Preparation of L-Aspartic acid

Input Mass (kg) Output Mass (kg)

L-Aspartic acid 319.4 Aqueous ZA solution 2185.5

Sodium Hydroxide 324 Waste 1261.5

Benzyloxycarbonyl

chloride 480

Toluene 1154

Water 1170

Sum 3447 Sum 3447

Preparation of DL-Phenylalanine

Input Mass (kg) Output Mass (kg)

DL-Phenylalanine 991.5 Aqueous DL-PM

solution 1007.1

Methanol 3320 Waste 6501

Hydrogen chloride 437.4

Sodium hydroxide 124.3

Water 2634.7

Sum 7508 Sum 7508.1

xiv

Condensation

Input Mass (kg) Output Mass (kg)

ZA aqueous solution 192.43 Z-APM.D-PM 1217

Aqueous DL-PM

solution 250.98

Thermolysin 7.2

Calcium acetate

monohydrate 1.3

Water 765

Sum 1217 Sum 1217

Purification

Input Mass (kg) Output Mass (kg)

Z-APM.D-PM 0.267 Aspartame 0.265

Hydrogen chloride 0.04

Ammonium formate 0.126

Methanol 2.4

Catalyst Pd/C 0.05

Water 28

Sum 30.88 Sum 30.62

Process and Systems Engineering Center (PROSYS)Department of Chemical and Biochemical EngineeringTechnical University of DenmarkSøltofts Plads, Building 229DK - 2800 Kgs. LyngbyDenmark Phone: +45 45 25 28 00Web: www.kt.dtu.dk/forskning/prosys